The Pentagon’s response, announced in January, was its first-ever direct equity investment in a defense supplier. A billion-dollar convertible preferred stake in L3Harris’s solid rocket motor business, ahead of L3Harris spinning that business off as a publicly traded company later this year. It is an unprecedented financial structure for the Department of Defense. It is also an open admission that the procurement system Washington has been using for forty years is no longer fast enough to fix the supply chain that builds those interceptors.

The investment is necessary, but it’s pointed at the wrong target.

A 10-Second Lesson on Rocket Science

Most missiles and interceptors in the U.S. arsenal run on what is called a solid rocket motor, or SRM for short. Unlike a liquid-fueled rocket, which mixes its propellant ingredients during flight, an SRM comes pre-loaded with a hardened block of fuel that simply ignites and burns. The design is simpler, more reliable, and shelf-stable for decades. You can sit an SRM in a silo or under the wing of a fighter for twenty years and it will still fire on command.

For the high-end use cases, this is irreplaceable. The submarine-launched nuclear deterrent runs on SRMs. The intercontinental ballistic missiles in silos across the Midwest run on SRMs. The interceptors designed to take down incoming ballistic and hypersonic threats run on SRMs. The country genuinely needs more of these, and the Pentagon is correct that the supply chain that builds them is too small for the strategic environment we are in.

What the equity check does not address is the larger problem hiding underneath: somewhere along the way, the same expensive, bespoke product class became the default answer for every job that needed a propulsion system. Including chasing $30,000 drones with $400,000 missiles.

How the Supply Chain Got This Brittle

At the prime level, two companies build solid rocket motors at scale in the United States: Northrop Grumman and L3Harris, which acquired Aerojet Rocketdyne in 2023. Though it may look like a two-supplier market, move one tier down and the competition disappears. Northrop and L3Harris share most of the same specialty suppliers, and many of the most critical inputs come from a single source.

The most striking example is the chemical that allows solid rocket fuel to actually burn, the ingredient that lets the fuel block release its energy in a controlled way rather than just sitting there inert. There is exactly one U.S. manufacturer of it, and the company is called AMPAC Fine Chemicals. It was acquired by NewMarket Corporation for roughly $700 million in 2024. Every Patriot, every Javelin, every Stinger, every Trident submarine missile runs on a chemical from that one company’s facility. NewMarket has announced a $100 million expansion, which helps, but does not change the structural reality. There is no second domestic supplier, and no certified import. The entire U.S. solid rocket motor industry runs through one chemical, made by one company.

The L3Harris deal is part of a broader Pentagon push to roughly 2.5x output across about a dozen priority weapons. Congress backed it with $6.4 billion in 2026 munitions procurement, including a dedicated $500 million for the SRM industrial base specifically and another $200 million in a separate reconciliation package. Lockheed Martin separately signed a seven-year agreement to expand Patriot interceptor production from about 600 units a year to 2,000. It’s more money flowing into this industrial base than at any point in a generation, and still it doesn’t solve the actual problem.

The Ferrari Problem

I’ve referenced this same issue in the turbine industry. The market has plenty of Ferrari-grade products. Bespoke, high-performance, low-volume, made by a small number of consolidated suppliers who have rational reasons not to expand. What the market does not have is enough of the simpler, cheaper, higher-volume alternatives that the next decade of demand actually needs.

The same logic applies to SRMs, just more extreme. A Ferrari is the right car when you need to win a race. It is the wrong car to deliver your uber eats order. We’re currently using Ferraris to deliver pizza, and we are running out of Ferraris.

Even if every dollar of the new funding worked perfectly, even if Northrop and L3Harris doubled output tomorrow, the United States would still be in a war of attrition we cannot win at scale. Doubling the production of $400,000 interceptors does not solve the underlying problem of needing to fire them at $30,000 drones. It doubles the burn rate.

The real constraint is not the size of the SRM supply chain, but rather that we have been pointing the most expensive, most bespoke propulsion technology in the world at threats that don’t require anything close to it. The fix is not just to make more Ferraris faster, the fix is to start fielding the propulsion equivalent of a Toyota.

Where the Real Opportunity Is

The companies building those Toyota-grade alternatives mostly don’t exist in the public markets yet. They are small teams that have spent the last two or three years quietly proving that a cheaper propulsion architecture can do 80% of the job for 5% of the cost. There are roughly three categories worth watching.

The first is attritable turbine engines for drones. I’ve mentioned this before and our portfolio has exposure to it. The argument is the same, for many strike, interdiction, and reconnaissance missions, a small, cheap, mass-produced turbine engine on an expendable airframe is a more rational answer than a missile carrying a high-grade SRM. The traditional primes are not optimized to build at this price point or this volume, which is exactly why the opening exists for new entrants.

The second is propeller-based propulsion, particularly for interceptor drones designed to take down other drones. A handful of companies are now building electric-propeller interceptors capable of engaging Shahed-class (Iranian Group 3 drones) threats at a unit cost an order of magnitude below traditional missile interceptors. The economics finally start to work. You are firing a cheap thing to kill a cheap thing, rather than firing a Patriot at a radio shack drone. I’ve personally got some exposure here as well.

The third is additive/advanced manufacturing for energetic materials. Think cheaper propellant to be used in cheap drones for short range and short lived systems. The qualification timelines for high-end defense applications remain real, and I’m not convinced yet that additive can handle the top of the stack. But for the lower end of the propulsion stack, where the chemistry is simpler and the tolerances are looser, additive methods can potentially decrease timelines and reduce dependence on the same sub-tier suppliers that bottleneck everything else.

The Strategic Bottom Line

At the end of the day, we need both. The high-end SRM supply chain is genuinely critical because hypersonic threats and ballistic missile defense and the nuclear deterrent are not problems a propeller-driven drone can solve. The Pentagon’s billion-dollar investment in L3Harris is not wrong, it’s necessary.

But the deeper investable shift is in the opposite direction. The next decade of defense capital will not be defined by who rebuilt the legacy supply chain fastest. It will be defined by who built the cheap, attritable, mass-produced propulsion systems that mean we are no longer firing $400,000 missiles at $30,000 drones in the first place.

The Ferrari is still useful when you actually need to win a race. Most of the time, what you really need is a Toyota.

Disclosure: These are my opinions, not investment advice. I have positions in the sectors I write about. My views are shaped by my investments, and my investments are shaped by my views. Take everything with appropriate skepticism and do your own diligence.

The Pentagon granted a national security waiver and deliveries resumed within weeks. Crisis averted, move on.

Except that was one magnet in one component of one aircraft. The real exposure is orders of magnitude larger. The same class of materials, rare earth elements, runs through virtually every advanced weapons system, every electric vehicle motor, every wind turbine generator, and most of the consumer electronics in your pocket. And the supply chain for nearly all of it runs through China. Not just the mining, the processing, the refining, the magnet manufacturing, and increasingly the rules governing who gets to buy the output. One country controls almost the entire value chain for materials that the modern economy cannot function without. That’s a dependency, not a supply chain risk.

What They Are and Who Controls Them

Rare earth elements are a group of 17 metals that, despite the name, aren’t actually rare. They’re found all over the world. What makes them “rare” in practice is that they almost never show up in concentrations high enough to mine profitably, and separating them from each other is extraordinarily difficult because all 17 exhibit similar chemical properties during separations. Think of trying to sort 17 nearly identical shades of gray by hand. You can do it, but it takes specialized equipment, deep expertise, and a lot of patience.

What makes them irreplaceable is their magnetic properties. The permanent magnets made from these elements are the strongest in the world, and they’re in everything: the motor in an electric vehicle, the generator in a wind turbine, the guidance system in a missile, the speakers in your phone. An F-35 contains roughly 920 pounds of rare earth materials. A single large wind turbine needs about 600 pounds of rare earth magnets. There is no viable substitute for most of these applications. Nothing else has the same properties.

In the same way that the Middle East has oil, China has rare earths. But unlike the oil producers who mostly drill and export crude, China did something far more strategic. Over three decades, Beijing built an entire ecosystem: mining, chemical separation, metal refining, alloy production, and magnet manufacturing, every step under state coordination, backed by billions in government investment. China accounts for roughly 70% of global mining, approximately 90% of processing, and over 90% of the world’s high-performance permanent magnets. They don’t just own the mine, they own every step between the mine and the finished product, and if you want to build anything that uses a permanent magnet, you’re going through them.

The part most people miss is that even countries that mine rare earths still depend on China to process them. The U.S. is the world’s second-largest rare earth miner, but until recently most of that ore was shipped to China for processing. We were mining these metals on American soil and sending them to our chief geopolitical rival to turn into something useful. China has filed more than double the rare earth patents the U.S. has, and the processing expertise itself has become a moat as deep as the mineral deposits.

When the Tap Gets Turned Off

China has used rare earths as leverage before, restricting exports to Japan during a maritime dispute in 2010, but 2025 made that look like a warmup.

In April, in direct retaliation for U.S. tariffs, China imposed export controls on seven rare earth elements critical to defense and energy applications, along with all related compounds, metals, and magnets. Export volumes fell sharply. Some automakers outside China were forced to cut production or temporarily shut down because they couldn’t source permanent magnets, and European prices for some elements hit six times the Chinese domestic price. Then in October, China escalated dramatically, adding five more elements and introducing a foreign direct product rule modeled on the same mechanism the U.S. uses to restrict semiconductor exports to China. The rule requires foreign companies to obtain a license from Beijing to export any product containing Chinese-sourced rare earth content, even products being shipped between two countries that aren’t China. Beijing watched how Washington used export controls to constrain China’s semiconductor industry and adapted the same playbook for a domain where China holds the leverage.

What’s most concerning about this is the escalation pattern. First raw materials, then processing technology, then jurisdiction over finished products, then restrictions on Chinese citizens sharing rare earth expertise overseas. A temporary trade truce in May 2025 was supposed to ease things, but Chinese authorities dragged their feet on licenses even while the truce was technically in effect. When your entire advanced manufacturing base depends on a single foreign supplier who has demonstrated willingness to turn the tap on and off, the word “risk” doesn’t capture it. It’s a structural vulnerability that’s being actively exploited.

What the U.S. Is Doing About It

Real money is finally moving, but I think most people underestimate how far behind we are.

MP Materials operates the only rare earth mine and processing site of scale in North America, and they’re building the first fully domestic mine-to-magnet supply chain. Their Texas facility began manufacturing permanent magnets in December 2025, making them the only company doing so on American soil. In July 2025, the DoD announced a multibillion-dollar partnership that went far beyond a typical procurement contract: the government purchased $400 million in equity, positioning itself as MP’s largest shareholder. It’s the first time the federal government has taken an equity stake in a critical minerals company. The deal includes a 10-year price floor and purchase commitment, and MP is building a second magnet facility to significantly scale domestic capacity. The government isn’t just placing orders anymore, it’s becoming a strategic partner, which represents a fundamentally different approach to rebuilding an industrial base that the traditional procurement model hollowed out.

Lynas Rare Earths, the Australian company that is the only commercial-scale rare earth producer outside China, has been building a processing facility in Texas with roughly $258 million in DoD funding. But the project has hit real headwinds, cost overruns and difficulty reaching acceptable terms have raised questions about whether it will proceed. Washington’s decision to go all-in on MP may have sidelined Lynas’s U.S. ambitions, which I think is a mistake. We need multiple sources, not a domestic monopoly replacing a foreign one.

Beyond those two, the U.S. signed an $8.5 billion critical minerals pact with Australia, struck agreements with Malaysia and Thailand, and backed a joint venture with Saudi Arabia’s national mining company to build a rare earth refinery in the Kingdom. The strategy is evolving from pure domestic onshoring toward building a network of allied processing capacity, which is probably the right approach. But building a new mine takes 10 to 15 years, processing capacity takes 5 to 7, and even optimistic analysts say a decade-plus of sustained investment is needed before we meaningfully reduce dependence on China. We are in the very early innings.

The Investment Lens

This is a long-duration structural problem with massive capital requirements, which means the companies solving it now will have enormous moats once they’re operational. That’s the thesis in a sentence.

For venture and private capital, the opportunities sit in three areas.

First, recycling and secondary recovery. The fastest path to rare earth supply without new mines is recovering these metals from end-of-life products and from waste streams at existing mining operations. That means pulling magnets out of old EV motors and turbines, but it also means using chemistry to recover valuable minerals from mine tailings and process solutions that are currently treated as waste. Apple invested $500 million in MP’s recycling capabilities, which tells you where the market is heading. One company I’ve personally invested in is developing a separation technology designed to selectively recover critical minerals from mining waste streams, essentially turning what used to be a disposal problem into domestic feedstock for refining. The economics are compelling because you skip the mining, the permitting, and most of the upstream processing.

Second, alternative processing technologies. If the bottleneck is the chemical separation expertise China has monopolized, then companies developing novel separation methods are solving the actual constraint. Qualification timelines are long and the science is demanding, which creates a natural barrier to entry that benefits first movers. This is where deep technical diligence matters most: does the team have the access, the talent, and the technology to actually deliver?

Third, domestic magnet manufacturing. China produces over 90% of the world’s high-performance permanent magnets, and even if the mining and processing problem gets solved, someone still has to turn refined metals into finished magnets. Global demand is projected to roughly triple by 2035, and current planned domestic capacity won’t come close to covering it. This is capital-intensive, infrastructure-scale investment, more suited to government partnership and private equity than venture, but there are opportunities in companies developing new magnet formulations that reduce dependence on the most supply-constrained inputs.

The Full Picture

The pattern here is the same one playing out across the industrial base: real demand, constrained supply, concentrated production in players who have rational reasons not to move faster. The CEO of one of the largest turbine manufacturers in the world is personally negotiating rare earth reserves with the U.S. government because their turbine blade coatings depend on it. That’s not a one-off. The turbine supply chain, the defense industrial base, and the energy transition all converge on this same set of materials, processed almost entirely in one country.

The mismatch that should keep investors up at night is the timeline. China can restrict exports tomorrow. The U.S. response, even moving at unprecedented speed with the government taking equity positions and striking billion-dollar allied mineral pacts, won’t deliver meaningful independence for another decade. That gap between how fast this problem is moving and how slow the solution can physically be built is itself an investment thesis, and the companies that are operational now or will be within the next two to three years aren’t just building businesses. They’re building the bridge that everything else has to cross.

Disclosure: These are my opinions, not investment advice. I have positions in several of the companies and sectors mentioned here. My views are shaped by my investments, and my investments are shaped by my views. Take everything with appropriate skepticism and do your own diligence.

Except it should have, because that single explosion disrupted production across more than 300 munitions programs for roughly two years. Javelins, Stingers, Hellfire missiles, artillery rounds, programs that sit at the center of U.S. and allied defense strategy, all stalled because one facility in rural Louisiana that nobody was tracking went offline. There was no backup supplier, no redundancy, just a crater where the supply chain used to be.

If you’ve read my thoughts on turbine supply chains, this pattern should feel familiar. The same structural fragility exists across the defense industrial base, and in many ways it’s worse. In turbines, the bottleneck is a handful of casting houses and forging specialists. In defense, the bottleneck is hundreds of small, undercapitalized suppliers scattered across middle America that the prime contractors (the Lockheeds, the Northrops, the Raytheons) often don’t even know exist.

That’s not an exaggeration. According to the Department of Defense’s own assessments, prime contractors have no visibility beyond their direct suppliers in roughly 84% of their programs. The companies responsible for building the weapons the U.S. military depends on cannot tell you who makes the raw materials or specialty components several levels below them in the chain, and that’s where the real vulnerability lives.

The Convergence Problem

The defense supply chain looks competitive at the top. Northrop Grumman and Aerojet Rocketdyne (now L3Harris after the acquisition) appear to be rivals in solid rocket motor production, and Lockheed and Raytheon compete for the same missile contracts. A quick rocket science lesson: solid rocket motors are the type of engine that uses a pre-packed solid fuel, as opposed to liquid-fueled engines that mix propellants on the fly. Most of the missiles in the U.S. arsenal, from handheld launchers to submarine-launched nuclear weapons, use solid rocket motors because they’re simpler, more reliable, and ready to fire at a moment’s notice. From the outside, the companies making them look like a functioning competitive market.

Go one or two tiers below and the competition disappears. Those “rival” programs share the same smaller manufacturers, the same specialty chemical producers, the same forging operations. If one of those shared suppliers goes down, it doesn’t take out one program, it takes out dozens across multiple prime contractors simultaneously.

AMPAC Fine Chemicals is the perfect illustration. AMPAC is the sole U.S. manufacturer of ammonium perchlorate, a chemical compound that acts as the oxidizer in solid rocket fuel, essentially the ingredient that allows the fuel to burn. Without it, solid rocket motors don’t work. Every rocket motor in the U.S. arsenal, from the Stinger and Javelin all the way up to Trident submarine-launched missiles, depends on this one chemical, and there is exactly one domestic company that makes it. If something happens to AMPAC, the United States cannot produce solid rocket motors, full stop, there is no second supplier to call.

The entire U.S. strategic and tactical missile inventory runs through a single chemical supplier in a single facility, and most investors in the defense sector couldn’t name the company if you asked.

The Chinese Alloy in Your F-35

The most expensive weapons program in human history, with a lifetime cost north of $1.7 trillion, had its entire delivery pipeline frozen in 2022 because of a magnet. A small magnet inside the F-35’s power system, made with a specialty alloy that turned out to have been sourced from China. Honeywell had subcontracted the work, and the Chinese material entered the supply chain somewhere below the company’s direct oversight. Nobody checked until someone did, and once they found it, every F-35 delivery stopped while the Pentagon investigated whether the material posed a security risk.

The Pentagon eventually granted a waiver and deliveries resumed, but the episode exposed something that should concern anyone investing in this space: foreign material dependencies aren’t being caught by design, they’re being discovered by accident. The Chinese material wasn’t the result of some deliberate intelligence operation, it got into the supply chain because a subcontractor’s subcontractor bought from the cheapest available source, which happened to be in China. That’s not a one-off, that’s how the entire system works.

Approximately 78% of U.S. weapons systems contain materials or components sourced from China. Rare earth elements get the most attention, China controls roughly 60% of global mining and over 85% of processing for these materials, but the dependency extends into specialty metals, electronic components, and chemicals that show up in everything from guidance systems to explosives. China announced new export controls on several of these critical materials in late 2023 and expanded them in 2024, including materials used in ammunition, night vision devices, and nuclear weapons components. When the country that supplies most of a critical input to your weapons decides to restrict it, that’s not a trade policy issue, that’s a national security emergency playing out in slow motion.

A Hollowed-Out Base

The Department of Defense lost approximately 40% of its small business suppliers over the past decade. That number explains almost everything else in this piece.

Small businesses make up the bulk of the lower tiers in the defense supply chain, producing the specialty components, raw materials, and niche manufactured parts that eventually end up in a missile or an aircraft. These aren’t companies with government affairs offices in Washington, they’re machine shops in Ohio, chemical producers in Texas, foundries in Pennsylvania, many of them with fewer than fifty employees.

They’ve been leaving the defense market for years, and the reasons are predictable if you understand the economics. Defense procurement is feast or famine. A program gets funded, orders spike, and the government expects suppliers to ramp production. Then the program gets cut or delayed in the next budget cycle, orders evaporate, and the supplier is left holding excess capacity they took on debt to build. After a few rounds of that, the owner of a 30-person machine shop looks at commercial aerospace or automotive, where demand is steadier and payment terms aren’t 90 to 120 days, and makes a perfectly rational decision to stop taking defense work.

On top of that, there’s the compliance burden. The maze of export controls, military procurement rules, and cybersecurity certifications that come with defense work impose real costs on small suppliers. A shop that wants to keep doing defense work now needs to invest in IT security infrastructure, maintain regulatory compliance programs, and navigate a procurement bureaucracy that was designed for billion-dollar prime contractors, not a company pulling in $5 million a year.

The suppliers that remain have tremendous pricing power but are simultaneously too small and too capital-constrained to scale when the government suddenly needs more product. When the Pentagon decided to ramp artillery shell production for Ukraine, the bottleneck wasn’t at General Dynamics or BAE Systems, it was at the smaller suppliers further down the chain who make the casings, the propellants, and the detonation components. Some were running single-shift operations in aging facilities because they couldn’t justify the capital expenditure to expand, and why would they? They’ve been burned before, margins are great in a shortage, and nobody has given them a reason to believe this cycle will be any different. Everyone in this ecosystem has a rational reason not to move faster, and that’s what makes the bottleneck structural, not temporary.

Where the Money Is Moving

There is hope! The Pentagon knows this is a problem, and for the first time in a while, real money is moving toward solutions.

The most significant signal came when the Department of Defense announced a roughly $1 billion equity investment alongside L3Harris to expand solid rocket motor production capacity. That’s not a typical procurement contract, it’s the government putting equity capital into industrial capacity, which is a fundamentally different approach than the traditional model of just placing orders and hoping the supply chain figures it out. The 2026 defense budget request included $6.4 billion specifically for munitions, a meaningful increase that reflects both the drawdown from Ukraine transfers and the recognition that production capacity needs to expand across the board.

On the private side, what I find most interesting is the amount of startups working on additive manufacturing (essentially advanced 3D printing for metal components) and other novel production techniques to bypass traditional bottleneck suppliers entirely. The logic is simple: if you can’t get a critical part from the two companies in the world that make it, and you can’t convince them to expand, then develop a manufacturing process that doesn’t need them at all. Some of these companies are printing metal components that previously required complex, multi-step production processes, compressing both the timeline and the supplier dependency.

I’m still not fully convinced that additive manufacturing is ready to produce critical defense components at the scale and reliability the military requires. The qualification timelines are long, the material science is still maturing, and proving consistency across thousands of parts is a very different challenge than printing a successful prototype. But the thesis, that you can re-architect the supply chain by eliminating tiers entirely, is compelling. The companies that interest me most aren’t the ones promising to 3D print an entire missile, they’re the ones identifying a specific bottleneck part that currently takes 18 months to procure from one of two suppliers and proving they can produce it through an alternative process in weeks.

The Investor Lens

The primes are well-covered by Wall Street, everyone knows Lockheed’s backlog and Northrop’s latest bomber program. But the real constraint, and therefore the real opportunity, is several tiers below in the companies that make the materials and components that everything else depends on.

The companies worth watching are the ones solving the actual bottleneck: domestic sources for critical materials currently dependent on China, novel manufacturing approaches that simplify the supply chain, and modular production systems that can scale without requiring billion-dollar facility investments. These aren’t the companies getting billion-dollar valuations on a pitch deck, they’re the companies that can actually deliver product when the primes come knocking, and increasingly that’s all that matters.

There’s a reason the Pentagon is now willing to put equity alongside companies rather than just placing orders. The traditional procurement model broke the supply base. Forty percent of small suppliers left. Single points of failure are everywhere. The fix requires patient capital, not purchase orders, and that creates an opening for venture and private equity that didn’t exist five years ago.

Before investing in the next defense prime or chasing the latest missile program, ask the same questions I always come back to: who actually makes the components? How many suppliers can do it? And what happens when one of them goes offline?

If you can’t answer those questions, you’re not investing in the defense supply chain. You’re investing in the logo at the top of it and hoping everything below holds together, and Minden already proved what happens when it doesn’t,

Disclosure: These are my opinions, not investment advice. I have positions in the sectors I write about. My views are shaped by my investments, and my investments are shaped by my views. Take everything with appropriate skepticism and do your own diligence.

The Department of Defense is shifting how it thinks about propulsion. Instead of engines designed to last 30 years and generate service revenue for decades, the military needs engines that are cheap, mass-produced, and expendable. Engines you throw away after a handful of missions. And the companies that have dominated jet propulsion for the last half century, the original equipment manufacturers (OEMs) like GE, Pratt & Whitney, and Rolls-Royce that design and build these engines, aren’t set up to deliver them. Not because they lack the talent, but because their entire business model is built around the opposite product.

The Shift

In February 2026, the Air Force awarded Collaborative Combat Aircraft (CCA) Increment 2 engine contracts to four teams: Beehive Industries, Honeywell with its SkyShot 1600, Pratt & Whitney, and a GE Aerospace/Kratos partnership with the GEK1500. These are engines in the 800 to 1,600 pound-thrust range, designed to power autonomous drones that fly alongside manned fighters.

The key word in every briefing, every contract document, every conversation I’ve had about this program: attritable. These drones aren’t designed to come home and get serviced for 20 years. They’re designed to be produced at volume, deployed in contested environments, and if they don’t come back, that’s part of the plan. If the airframe is expendable, the engine has to be too.

The Air Force has been explicit about what it needs: high-performing, low-cost engines that enable the disruptive capabilities of small CCAs. That language matters. “Low-cost” and “jet engine” have never really coexisted in aerospace, the entire industry was built on the opposite premise.

The Business Model Problem

The OEMs don’t make their money selling engines. They make their money maintaining them. A traditional jet engine might sell at thin margins or even a loss, but the installed base generates service contracts at 50%+ margins for decades. The engine gets you in the door, the maintenance agreement is where the real economics live. That model has worked incredibly well for a long time.

Now ask those same companies to build an engine that’s designed to be thrown away. No aftermarket, no service contracts, no decades-long revenue stream per unit. Just hardware, sold at the lowest possible cost, in the highest possible volume. It’s like asking Ferrari to build disposable vehicles. The engineering talent is there, but every incentive in the business pushes in the opposite direction.

A traditional jet engine has thousands of precision-manufactured parts that require investment casting single crystal superalloys, and years of development and certification (if those sound complicated, it’s because they are). Every one of those processes is optimized for durability, because durability is what keeps the customer coming back to your shop every few thousand flight hours and paying for the privilege. When that’s your business, you design for expensive and permanent, not cheap and fast. The incumbents will compete for attritable contracts because the dollars are real, but their DNA, their supply chains, and their margin expectations all pull them toward what they’ve always done. That’s the opening for everyone else.

Rethinking How Engines Get Built

If the demand is for cheap, high-volume, expendable engines, then the way engines have traditionally been manufactured doesn’t work either. You can’t hand-assemble thousands of precision parts through a supply chain designed for low-volume, high-margin production and expect to hit the price points the military is asking for. Something has to change in how these things actually get built.

That’s where a lot of the innovation is happening right now. Part consolidation, where hundreds of individual components get designed into a handful of integrated pieces. Simplified supply chains that don’t depend on the same two or three specialty casting houses that the entire aerospace industry shares. Faster iteration cycles, where a new engine design can go from concept to test in months instead of years. Some companies are using additive manufacturing to get there, others are streamlining quality control, and more, but the direction is the same: strip out the complexity that was built for durability and replace it with production methods built for volume and cost.

Several companies in this space have already completed fire tests and high-altitude testing with engines built through these newer approaches. This isn’t theoretical, hardware is running and performing. The question isn’t whether you can build a capable jet engine this way. The question is who scales it first.

Where the Money Is

The UAV propulsion market sits at roughly $7 billion today and is projected to reach somewhere between $11 and $19 billion by the early 2030s, depending on whose estimates you trust. The military segment dominates, but commercial drone applications are growing fast across surveillance, logistics, agriculture, and infrastructure inspection. As payloads get heavier and missions get longer, battery-electric won’t cut it for a lot of these use cases.

A company in our portfolio builds turbine engines for drone applications. What drew us to the company, other than an incredibly capable team, was that the economics of disposable engines fundamentally favor a different kind of manufacturer. The primes will win some contracts because they always do, they have the relationships, the lobbying infrastructure, and the balance sheets. But the startups that can build at volume, at lower cost, with production approaches the incumbents weren’t designed around, those are the ones building real businesses. Not just collecting development contracts.

Same Thesis, Different Altitude

The military has decided it needs thousands of cheap, expendable autonomous aircraft, and the engines that power them need to be produced at a pace and price point that the traditional aerospace supply chain was never built for. When everyone else is focused on who wins the prime contract, pay attention to who can actually deliver. That’s where the value is.

Disclosure: These are my opinions, not investment advice. I have positions in the sectors I write about. My views are shaped by my investments, and my investments are shaped by my views. Take everything with appropriate skepticism and do your own diligence.

But let’s be honest about what it proves. Running one GPU on one satellite is a proof of concept for computing in space, but it is not evidence that orbital data centers can compete with what’s on the ground. And the company behind it just hit a billion-dollar valuation 17 months after demo day. They’re not the only ones chasing this idea. SpaceX filed plans for up to a million orbital data center satellites. Google announced a space AI effort. Blue Origin is exploring similar concepts.

The pitch is seductive: unlimited solar power, no permitting fights, no grid constraints. And there are real uses for compute in space. But the version of this story where orbital data centers replace terrestrial ones runs into three problems that money can’t solve.

You Can’t Cool a GPU in a Vacuum

People already know that space is cold, but they don’t realize that when it comes to actually cooling hardware, that doesn’t help you.

On Earth, data centers cool their hardware with air and water. Fans blow, coolant flows, heat moves. In space, there’s no air and no water. The only way to get rid of heat is to radiate it, the same way a hot stove glows. The problem is that at the temperatures GPUs run at, radiation is incredibly slow. A surface the size of a desk can only radiate about 250 watts. A single AI chip generates more heat than that in a package smaller than a shoebox.

Andrew McCalip, a space engineer at Varda Space Industries, built the most detailed public model of orbital data center economics. His code is open source and his assumptions are transparent. His thermal analysis shows that an orbital GPU would run with only about 12 degrees of margin before it starts overheating and throttling performance. That’s razor thin. And the radiators you’d need to cool everything are roughly the same size as the solar panels powering everything, meaning you’re basically doubling the size and mass of every spacecraft just for cooling.

You can engineer around a lot of things. You can’t engineer around the basic physics of how heat moves in a vacuum.

The Math Doesn’t Pencil

McCalip’s model compares the full cost of running data centers in orbit versus on the ground. No hand-waving, just math based on publicly available numbers.

The result: a 1 GW orbital data center would cost roughly $51 billion over five years. A comparable ground based facility costs about $16 billion. That’s a 3x cost disadvantage, and that’s with aggressive assumptions on launch costs and hardware efficiency, and before you factor in that terrestrial gas turbines generate power at a fraction of the cost of solar panels bolted to satellites.

The counterargument is always that launch costs are dropping. True, but even Starship at $500 per kilogram doesn’t close the gap. The problem isn’t any single cost. It’s that you have to buy the rocket, the spacecraft, the power hardware, the deployment, and pay a margin at every step. Those costs stack on top of each other. The only way to make the math work is if one company controls the entire chain from rocket to GPU. That’s not a startup, that’s SpaceX, and even then it’s a stretch.

Radiation, Bandwidth, and the Downlink Problem

GPUs already fail at a rate of ~9% per year in normal data centers. In orbit, cosmic rays hit chips and corrupt calculations. Smaller, more powerful chips are actually more vulnerable, not less. You can build protection against this, but it adds weight, cost, and slows everything down. Running one chip for a demo is a very different proposition than running thousands around the clock for years.

Then there’s the bandwidth problem. A satellite in low orbit can only talk to a ground station for brief windows as it passes overhead. The best optical downlinks demonstrated by NASA can achieve 200 Gbps. That is fast, but it only works during those short windows. Meanwhile, a data center on the ground moves more data internally in one second than an orbital system could downlink in a day. Any workload that needs real-time response, which includes most of what AI is actually used for, doesn’t work with the delay of sending data up and results back down.

Where Compute in Space Actually Makes Sense

None of this means putting processors in orbit is pointless. It means the use case is fundamentally different from what’s being sold.

If you have a satellite taking pictures of Earth, processing those images on the satellite before sending them down makes enormous sense. You can cut the amount of data you need to transmit by up to 90%. Real-time wildfire detection, maritime monitoring, missile tracking, these are applications where having a processor next to the sensor is genuinely valuable.

That’s edge computing. Small, purpose-built processors sitting next to the thing they’re processing data for. It’s a real market and a real business. But it’s a fundamentally different product than “we’re building data centers in space to solve the AI energy crisis,” which is the pitch that’s attracting billion-dollar valuations right now.

Running one GPU on one satellite is actually perfect validation for edge computing. It proves you can process data locally in orbit. Scaling that to a constellation of thousands of satellites running coordinated AI training is a completely different problem, and it’s the problem that physics doesn’t cooperate with.

The Investor Lens

I’m not rooting against anyone in this space. If companies can build real businesses around space-based edge computing, that’s valuable and I’d consider investing in the right team. But the valuations floating around right now aren’t priced for edge computing, they’re priced for the vision of replacing terrestrial data centers, and that vision doesn’t survive contact with the math.

When Microsoft’s president was asked about orbital data centers, he said they might eventually pursue the strategy but for now they’re “keeping their feet on the ground.” I believe that’s the right instinct. The companies actually solving the compute power crisis are doing it with terrestrial infrastructure. Gas turbines, modular power, behind-the-meter generation. It’s not as exciting as launching GPUs into orbit, but then again it works.

Before you invest because the vision sounds transformative, ask yourself: does the math work at the scale they’re describing? Have the fundamental constraints been addressed, or just acknowledged? And is the valuation pricing the business they can build, or the business they’re pitching?

Edge compute in space is real. Orbital data centers, as currently pitched, are not. Knowing the difference is the whole job.

Disclosure: These are my opinions, not investment advice. I have positions in the sectors I write about. My views are shaped by my investments, and my investments are shaped by my views. Take everything with appropriate skepticism and do your own diligence.

Most people hear backlog and think problem, but venture heard it and saw an opportunity so obvious the market might as well have been screaming it.

Turbines, in every form factor from powering a data center to powering a drone, are having a moment. And too many investors are either asleep or looking at the wrong part of the value chain.

The Numbers That Matter

GE Vernova’s total backlog totaled $150 billion by the end of 2025. Gas turbine equipment orders and slot reservations grew from 62 GW to 83 GW in a single quarter. The slot reservation agreements alone, essentially prepayments to secure future production capacity, jumped from 29 GW to 43 GW. Pricing on those new slots is running 10 to 20 percent above existing backlog. When customers are voluntarily paying a premium just to get in line, you don’t have a demand question. You have a supply question.

The U.S. gas turbine order pipeline went from 7 GW in 2020 to 44 GW in 2025. That’s 6x in five years. NextEra’s CEO said turbine prices have tripled in the last 24 months. An EPC contractor trying to staff power projects has to hire six thousand people to retain a thousand, because everyone’s losing workers to data center builds, semiconductor fabs, and LNG terminals.

This isn’t cyclical. This is structural.

Why It’s Stuck

Dylan Morris, an Founder & Engineer whose Turbine Tuesday series is one of the best resources on this industry, wrote a detailed breakdown of the turbine supply chain that anyone investing in this space should read. I’ll give you the investor version.

An F-class gas turbine has roughly 25,000 parts. The OEMs make almost nothing in-house. GE Vernova has 12,000 suppliers. Siemens has 40,000. Most of the supply chain, the machining, raw materials, specialty processes, can scale with money and time. But the strategic services, things like investment casting, single crystal casting, and forging, make up over 80% of the cost of a turbine. And there are maybe two to four suppliers in the world for each of those processes.

Here’s the part that matters for investors: those suppliers aren’t incentivized to scale. They’re publicly traded, margins are great in a shortage, and they’ve been burned by boom-bust cycles before. Why pour billions into new capacity when you’re already pulling 20%+ margins on constrained supply? The OEMs themselves don’t even make money selling turbines. They give all the margin to the supply chain and make it back on service contracts at 50%+ margins. Everyone in this ecosystem has a rational reason not to move faster. That’s what makes the bottleneck structural, not temporary.

Then layer on the raw materials. Seventy-five percent of titanium comes from Russia or China. Nickel superalloys require exotic elements mined in a handful of places globally. GE Vernova’s CEO is working directly with the U.S. government to secure yttrium reserves for turbine coatings and reduce dependency on China. When the CEO of a $100 billion company is personally negotiating rare earth supply with the federal government, you’re not dealing with a short-term hiccup.

Hyperscalers Can’t Wait

Data centers now drive roughly half of projected U.S. power demand growth through 2030. Electricity demand is growing at 2% annually, double the prior decade. And the companies building these data centers have a simple, expensive problem: they need power now, the grid can’t deliver it, and every month of delay costs them billions in lost AI compute revenue.

So they’re going behind the meter. Building their own generation, bypassing the grid entirely.

One analysis identified 46 data centers planning their own gas plants totaling 56 GW, roughly 30% of total planned U.S. data center capacity. Meta deployed 813 modular micro-generators at its El Paso facility. A company in our portfolio builds modular natural gas turbine systems designed specifically for this problem. They’re pre-assembled units that ship fast and deploy faster than conventional alternatives, and can scale from single units to full plants as demand grows. That’s the kind of product the market is desperate for right now. Some OEMs are reportedly shipping turbines without rotors or blades and installing them on-site later, just to hold delivery schedules. When you’re shipping a turbine without the parts that make it a turbine, that tells you everything about where demand stands.

Most of these deals flow through utilities and developers, not directly from hyperscaler to OEM. NextEra has over 20 GW of gas in its pipeline and expects to serve 30 GW of data center load by 2035. But the signal is unmistakable. The biggest, most well-capitalized companies in the world have decided that gas turbines are the fastest path to the power they need. Not solar. Not batteries. Not nuclear. Turbines.

Turbines in the Sky

The same supply-demand imbalance is playing out in defense, just at a different scale.

The military’s Collaborative Combat Aircraft program, autonomous drones designed to fly alongside manned fighters, awarded engine contracts in February 2026 to Honeywell and a GE Aerospace/Kratos team. These are 800 to 1,600 pound-thrust turbine engines, but built for a completely different philosophy: attritable. Designed to be low-cost, produced at volume, and expendable. Not maintained for decades like a traditional jet engine.

This matters because the incumbents aren’t built for it. Traditional aerospace engine manufacturers optimize for longevity and service revenue, the same dynamic as industrial gas turbines. Attritable programs need speed and production scale. It’s the same pattern: real demand, constrained supply, and a form factor that existing players aren’t set up to deliver. The UAV propulsion market sits at roughly $7 billion today and is projected to reach $11 billion by 2030.

A company in our portfolio builds turbine engines for drone applications. What drew us to the deal was exactly this pattern. When demand outstrips supply and the incumbents have structural reasons not to move fast enough, the companies that can actually deliver product capture outsized value. That’s true whether the turbine powers a data center or an autonomous wingman.

The Full Menu

Turbines aren’t the only answer to the power problem. I’m genuinely optimistic about several technologies that will matter over the next decade and beyond.

Nuclear will be part of the mix, and we’re backing teams we believe will get there. But the timelines are still not fast enough to satisfy our needs. TerraPower is targeting 2031. Holtec is looking at the mid-2030s. Kairos has a demo planned for 2029. The right teams with the right access and the right technology will eventually deliver. But “eventually” doesn’t solve the power crisis that hyperscalers, utilities, and industrial operators are staring at right now.

Geothermal is advancing. Fervo Energy has cut drilling costs in half and is targeting 500 MW by 2029. Promising, but still developer projections, not delivered power.

Gas turbines are the only firm, dispatchable power source that can be deployed at the scale and speed the market requires this decade. And when nuclear and geothermal do come online, they won’t replace turbines. They’ll join them. The demand curve is steep enough that we’ll need everything. Turbines are both the bridge to the rest of the menu and a permanent item on it.

Where the Money Is

Here’s how I think about it.

GE Vernova’s stock is up over 430% since its March 2024 spinoff. Siemens Energy’s profits nearly tripled in early 2026. But if you’ve been paying attention to the supply chain dynamics, the OEMs are giving most of their margin away. The real pricing power sits with the consolidated suppliers: the casting houses, the forgers, the coating specialists. And with the startups building turbines for form factors the incumbents aren’t optimized to serve.

We have investments across this space. Modular power generation for data centers and industrial applications. Attritable turbine engines for defense. Two very different markets, same thesis: turbine demand is structural, supply is constrained, and the companies that can actually deliver product will have pricing power most industrial businesses only dream about.

When you see a seven-year backlog, prices tripling, and the most capitalized companies in the world scrambling to secure supply, that’s not a sector with problems. That’s a sector in breakout. The constraint isn’t demand. It never was. The constraint is the ability to deliver.

Follow the backlog, not the headlines. Talk to the suppliers, not just the OEMs. And when everyone else is fixated on the technology that might work in 2035, pay attention to the one that’s sold out through 2032.

Disclosure: These are my opinions, not investment advice. I have positions in the sectors I write about. My views are shaped by my investments, and my investments are shaped by my views. Take everything with appropriate skepticism and do your own diligence.

Honestly, it’s very difficult. I use a mix of organizational tools, constant reminders, CRMs and I still fall behind. Nonetheless, I try to spend time every day checking in with others, because keeping up with people isn’t a tactic, it’s the foundation of everything I’ve built.

My entire career exists because of relationships, some of which were built before I was even born. My partners at the fund, the founders I’ve invested in, my investors, every single one started with someone I got to know and trusted who decided to take a chance on me. For a lot of people it wasn’t about my pitch or track record, it was because they knew me, they knew my family, and they wanted to help someone from their community succeed.

I’m self aware enough to know that when I started out, half the reason my earliest supporters were even willing to talk to me was because of the love and respect they hold for my father. He built a reputation as a stern but fair businessman, he respected the people he worked with and in turn received that same respect. I learned from him to negotiate hard, but never let it go so far that the other person walked away not wanting to work with you again. People remember that, and it’s because of his reputation that when I showed up years later with an opportunity and a vision, they saw those same values and decided to hear me out. I will be forever grateful for that.

But having that foundation doesn’t mean I had it figured out. It took time to learn that it’s one thing to meet someone, it’s another to know them, and the latter matters infinitely more.

Hands open, not out

Early on, I walked into plenty of meetings with nothing but asks. I’d have a pitch, a request, and a hope they’d give me what I wanted, and more often than not, I’d get a polite smile and never hear back.

Without naming names, I once found myself sitting for coffee with the owner of an NBA team. He’s a successful businessman, and so I went in doing exactly what I always did, pitching at full speed, throwing everything I had at him. He stopped me mid-sentence and said, “Look, whatever you’re doing now, this doesn’t work for me. You’re talking really fast, you’re jumping all around and honestly I can’t tell if any of this is real or not because of it.”

I think it came off more harsh than he intended but it was some of the best brutal feedback I’d ever received. I stopped speaking mid-pitch, apologized, and said I appreciate the feedback. I shifted, slowed down, and finished what I was saying. As we were leaving he actually apologized fearing he’d been too blunt, which I waved off. He then did something I didn’t expect, right before we wrapped up he said “Before we go, if there was any one thing I can do to help you what would it be?”

I was taken aback. Here was one of the wealthiest most influential men in the country, who more or less just endured a terrible pitch that he himself thought to be nonsense. I thought about it for a moment and realized he had already given me something invaluable. I told him no need, that his advice was some of the most brutal yet honest and useful feedback I’d ever received. Too many people I’d dealt with were afraid to be offensive, but he wasn’t and I’m grateful for that.

What I took away from that conversation changed how I approach every room I walk into. Not every person should be spoken to the same way. Some people process information fast, some need a slower pace. I needed to connect with who I was speaking to first and tailor the conversation to them.

They’re all just people

That shift, learning to actually connect with people instead of pitch at them, opened doors I never expected. Through my (sometimes excessive) talkative nature, I’ve found myself in conversations with CEOs, billionaires, fund managers, and people with influence I never could have imagined being in a room with, at the end of the day, they’re all just people.

I remember once getting a chance to speak with an extremely successful Persian entrepreneur. He had founded multiple unicorn companies, one of which had already IPO’d and there was a line of people waiting to talk to him. I (perhaps obliviously) went right up to him and started a conversation. I was so thrilled to see another Persian doing things in the same sector as me that the conversation ended up having nothing to do with business. We talked about culture, his family, his career. How he wishes he taught his kids more Farsi and how hard it was starting out when he immigrated. We connected over parts of our shared experience, and I think it was one of the longer conversations he had that day, probably because it was one of the only ones that had nothing to do with business or money. It was just two people connecting.

People at that level are bombarded 24/7 by others coming to them with their hand out, talking about work, asking for something. If you walk in trying to genuinely connect with them as a person, you might find a way to both win, and make a friend in the process. He gave me his personal number and even connected me with the person who runs his family office to talk business, and I still do my best to keep up with them both whenever they have the time.

Value is a two-way street

The amount of funds I’ve spoken with, shared deal flow with, and collaborated with openly has grown exponentially since I stopped treating information like something to guard. I’ve watched too many investors try to gatekeep introductions and hoard information, and for what? Sharing info is how I got an edge starting out, and it’s part of how I keep an edge every day. That information can be companies, investors, government programs, whatever is useful, and it all adds up.

It always comes back around. Funds I’ve shared deal flow with now refer me to potential investors. People I’ve introduced to one another remember the connection and continue to reciprocate even years later. Value is a two-way street, people rarely do things for free, and they shouldn’t have to.

The system behind it

It’s not easy, and you have to find your own system that works. For me, I have a list of everyone I talk to for work (a CRM for any tech readers). The key is I try to log whenever I last spoke to someone to make sure it never goes too long between seeing or talking to them. I gratefully have a good memory, but I also try to keep notes on people, their lives, their families, notable life events. That way whenever I talk to them I can pick up wherever we left off.

To me, a check-in should be a call or an in-person meeting, not a text. Sending someone an article or a quick message doesn’t build a relationship. You have to actually speak to each other, and even better if you can do it face to face. Reach out by text or email first and find a time that works for them. People are busy, and it’s better to work around their calendar first and yours second. Just checking in for the sake of caring goes a long way.

The longest game

These relationships are meant to last decades, not deals. There are engineers I met when I was starting out where I just tried to be helpful, get to know them, connect them with people that could help. Many of those same engineers are now executives of massive startups, and often try to reciprocate now too. I persistently check in, try to meet up for coffee or lunch whenever their schedule permits, and just like back then try to be helpful however I can. When we catch up it’s about family or hobbies or fun new restaurants, not just business.

The best part is that since I love what I do, these check-ins aren’t obligations, they’re friendships. They’re a chance to talk to someone else who is just as passionate as you are and nerd out about a cool topic.

Those dinners, drinks, social events, catch-up calls, seeing how families are doing, making introductions just to be helpful, it all speaks to who you are. I’ve had people reach out to me because they heard someone else speaking highly of me, wanting to invest or work together because of it. That is and always will be my goal. Deals last years, but your reputation lasts a lifetime.

Disclosure: These are my opinions, not investment advice. I have positions in the sectors I write about. My views are shaped by my investments, and my investments are shaped by my views. Take everything with appropriate skepticism and do your own diligence.

For real estate investors this is second nature, almost every investor I know in the industry deals with taxes in mind from day one. They offset income by depreciating their properties and when it’s time to sell, conduct something called a 1031 exchange. In that process you buy a similar property and can defer the tax you’d normally owe on the increased value. You can keep kicking that can down the road and, if you’re disciplined, theoretically never pay that tax. It’s not a loophole, it’s an understanding of how the system was designed and using it to your advantage.

Unfortunately, venture doesn’t exactly have the same tax structure benefits but that doesn’t mean there aren’t other structures that can help. Let me be clear, there isn’t a built-in mechanism to defer or eliminate capital gains from venture investing UNLESS the entity you invest from or the company you invest in are already exempt.

These are structures I personally utilize, look for, or even prioritize when evaluating deals, and can be the difference between saving a huge chunk of your gains or making a hefty donation to the IRS.

Self-Directed IRAs

For anyone who doesn’t know, IRA stands for Individual Retirement Account, kind of like a 401k. A Roth IRA is the same thing but the account is funded with post tax dollars. For most people, these accounts are a means of investing in the stock market and maintaining conservative growth, but they can be so much more.

A quick 101 on the Roth vs Traditional IRA structure:

In a Traditional IRA:

Money comes out of your income pre-tax → you invest → the money grows → you can pull the money out around the age of 60 and pay taxes on whatever you made.

In a Roth IRA its the same but flipped:

You put money in after paying taxes → you invest → the money grows → you can pull the money out around the age of 60 tax free.

Once you hit retirement age use it as a free bank account and enjoy.

The interesting thing is that these accounts aren’t just limited to investing in the stock market. Retirement accounts have “custodians,” usually a big brokerage like Charles Schwab, Fidelity, or ETrade. They’re basically just there to ensure you’re not doing anything illegal, but it also typically limits the account to just the stock market. However, if you work with a self-directed custodian (there are lots to choose from), they act as the middle man between you and the investment you’re making to make sure everything’s being done kosher. Now you can invest into private companies the same way you did stocks, all while limiting your tax exposure.

This isn’t exactly new information, Peter Thiel is the one who famously did this. Early on during his PayPal days, he used a self-directed Roth IRA to purchase early PayPal shares at a fraction of a cent per share, putting in roughly $1,700, the maximum contribution at the time. He flipped those into a huge profit and recycled that same strategy into his other portfolio companies Facebook and Palantir. Between the success of those companies and his tax free compounding, his Roth IRA is reportedly worth as much as $5 billion, all tax free as long as he waits till he’s 60 to pull money out.

When Thiel first started, there was a contribution limit of $1,700 per year, but as of 2025 the limit is $7,000 and slated to go up to $7,500 for 2026 if you’re under 50. One big caveat: you need to be making less than $150K per year ($250K jointly if you’re married). If you’re over that threshold, first off, good for you. Secondly, there’s still backdoor ways to invest in a Roth, basically through contributing to a regular IRA, paying tax and then converting it to a Roth. Check with your CPA of course but this opens the door for almost anyone to invest into a Roth to have true tax free growth.

Once it’s in a Roth, setting up a self-directed account isn’t super difficult, but unfortunately most of the custodians are. Most of them are a bit clunky, paperwork is slow and often done by hand, and no one seems to have a good sense of what is going on. If by some stroke of fate you find one that’s truly seamless and easy to use, please reach out and tell me.

Something important to note: you can’t borrow against your Roth IRA or use it as collateral for a loan. Unfortunately this means whatever money you make in that account really doesn’t get to be enjoyed until retirement or else you risk disqualifying the account. On the one hand you can’t touch the money, but on the other hand that’s the point, you can compound infinitely without taxes.

With that in mind, you should be using your Roth for your highest-conviction and fastest-returning deals that you expect will generate returns the quickest, then recycle that money into the next deal and the next and so on until you’re Peter Thiel.

QSBS

A Roth IRA isn’t the only tax advantaged strategy investors should be utilizing, the other is investing in QSBS companies (Qualified Small Business Stock).

For the more technical readers: under Section 1202 of the tax code, if you invest in a qualified small business, a domestic C-corp with aggregate gross assets of $75 million or less at the time of your investment, and you hold that stock for at least five years, you can exclude up to 100% of your capital gains from federal taxes.

In simpler terms, when investing in startups, if a company has less than $75M in total assets (or in many venture companies this can be less than ~$75M in total raised capital), then you don’t have to pay capital gains taxes federally as long as you hold for five years. Now there is a limit because there’s no such thing as a truly free lunch, but it’s a pretty high one. The tax exclusion is capped at the greater of $15M or 10x your original investment, per company. That means the benefits stack across multiple investments.

Using real numbers, if you invest $100,000 into a company that qualifies, and after five years that investment is now worth $2M, the $1.9M gain is (federally) tax-free. Even better, in states with no income tax like Texas or Florida that means you keep every cent.

One important detail: QSBS only applies to shares acquired directly from the company at original issuance, so secondary purchases don’t qualify.

But wait, there’s more! The One Big Beautiful Bill Act added another layer for tiered holding as long as the investment was made after July 4, 2025. This means that you don’t have to hold the investments for the full 5 years to get some tax benefits. If you only hold the investment for three or four years, you can still exclude 50% or 75% of your gains from federal taxes. For anyone complaining about a 3-4 year hold, that’s pretty aggressive for venture anyway, and if for some reason you’re cashing out before that then good for you, teach us your ways, and don’t worry there’s still hope for your taxes too.

There’s a 1031 equivalent for QSBS, referred to as a 1045. As long as you held for at least six months, you can reinvest the proceeds into another qualified small business within 60 days and defer the gain entirely. It’s not identical to a 1031, but the principle is the same, and if for any reason you cashed out within 6 months, again good for you, but this time just suck it up and eat the tax, you’re doing fine.

Of course there are limitations, so check with your CPA or your favorite AI (but still preferably your CPA). Also check with your local state laws, unfortunately for my fellow Californians, the state doesn’t like it when you save money on taxes, so you still pay state tax on QSBS gains (as much as 13.3%) even if it is completely excluded federally. In real numbers, if somehow you hit the top threshold and have $15M in gain, that means ~$2M goes to California and the next three inches of a high-speed rail line our great grandkids would be lucky to use. That’s money that someone in Texas or Florida would have just kept.

Still it’s worthwhile to look for QSBS opportunities, depending on where you live, you can save anywhere from 10% to 23.8% on your taxes. Let me be clear, whether a company is QSBS or not shouldn’t dictate whether you decide to invest, but it certainly helps.

The Bigger Point

I’m not a tax attorney, and this isn’t tax advice, but the point remains: if there’s money to be saved you should be saving it. You should be utilizing one or both of these structures in your investments, or be prepared to pay a lot more in tax than you need to.

When looking at returns, make sure you’re looking at net returns, not just net of fees, but what actually goes into your bank account at the end of the day. A deal that returns 5x but is fully taxable may end up making you less than a deal that returns 4x inside a Roth or if it was QSBS. The math changes when you account for what the government takes, and that math should be part of your process from day one, not something you figure out after you’ve already made money.

Real estate investors figured this out decades ago, it’s time venture investors caught up.

Disclosure: These are my opinions, not investment advice. I have positions in the sectors I write about. My views are shaped by my investments, and my investments are shaped by my views. Take everything with appropriate skepticism and do your own diligence.

That single concept has shaped more of my investment decisions than any financial model ever could, but I still feel like it’s lost on most investors. Too many people evaluate opportunities as standalone ones. Yes, it matters whether the concept has solid fundamentals, or if the valuation is favorable, or if there’s a strong team, but arguably the first question you should be asking is more basic: does the risk and timeline make sense relative to your liquidity (cash) needs?

Your investments aren’t separate buckets, it’s all part of a whole. Your house, stocks, income, venture positions, cash, debt, all of it should play a factor in how you make investment decisions. Those other buckets determine how liquid you truly are and in turn should help you determine your risk tolerance. You need to take it all into consideration when deciding whether or not to write a check.

Where Everything Sits

Every asset class sits somewhere on a spectrum from liquid and stable to illiquid and volatile. More or less, that spectrum goes like this:

Gov. Debt/Bonds → Real Estate → Private Equity → Stocks/Public Markets → Venture Capital → Caesars Palace

Venture and the casino are closer than most people would care to admit. The main difference being that venture usually has better odds, and unlike the casino, you have to wait years to find out if you’ve won.

Still the safer asset classes have their own drawbacks. It’s very hard to get rich off bonds, and real estate, while an excellent source of steady cash flow, usually won’t return you multiples on your money. Private equity requires you to be very hands on or trust someone else to handle your money and charge you for the privilege. Stocks mean that at any given time someone influential (President/South African Billionaire) can tweet something and dramatically change the value of your investments. Venture can make you huge multiples, but betting wrong can mean it all goes to zero. And a casino is pretty similar to venture but you can usually get a few free drinks out of it too.

A healthy portfolio would hopefully encompass a mix across these asset classes, and take risk tolerance into consideration. The question isn’t just “is this a good investment?” It’s “does this investment fit where I am and where I’m trying to go?”

The Wake-Up Call

I learned that risk is in every asset the hard way. In 2020, pre-COVID, my family’s investment portfolio was more or less all real estate, so when COVID hit, we got hit disproportionately hard. That was the moment I realized being diversified within an asset class isn’t the same as being diversified across them.

Since then, we’ve been working toward a better balance. Our goal is to still maintain our initial stability with real estate as the foundation, but now have meaningful positions in venture, public markets, private equity, and debt. Ideally, we want to build a portfolio that doesn’t get devastated by any single market cycle or occurrence. That shift may sound simple, but it’s also meant passing on plenty of objectively good opportunities because it didn’t make sense for our broader goals.

Who Should Be in Venture (and Why They’re Not)

Having started in real estate, I spend a lot of time with real estate investors. They’re usually the least risk tolerant investors around, and for good reason. They’ve built their lives slow and steady on assets growing ~15% a year and managed to offset almost all income tax in the process. It’s a good life, but it takes a long time to build.

Venture is almost the polar opposite. No cash flow, minimal control, a real chance of losing everything, all on similar timelines to real estate. Even worse, at least if a real estate deal goes south you can fire sale it and recoup some of your money, but in venture, it usually means that money is lost.

And yet real estate investors are arguably some of the best candidates for venture. They’re patient, they understand compounding, they’re used to limited liquidity, and they recognize that some investments require additional capital to gain full value. The struggle is that they like to manage their own capital and want cash flow, both things that are difficult to have in venture if not impossible.

What Founders Miss About Their Investors

Liquidity isn’t just an investor problem, founders deal with it too. That tension extends through everyone involved in the company.

While many founders understand that most of their investors are using OPM (other people’s money), they don’t necessarily know the restrictions that come with that. Fund managers have fiduciary responsibility to return capital within certain timeframes typically 10 years at a time, but independently to survive and raise their next fund, they need to start returning money as soon as 5-7 years in and showing huge markups. That can pose a real problem for early stage investors since you don’t always get to an IPO or acquisition in 5-7 years let alone 10.

It’s important for founders to realize this because when a fund manager is asking to sell out of a position, it isn’t a betrayal, it’s responsibility. More often than not they don’t have much choice in the matter, but a little leniency here can go a long way.

If founders allow for more flexibility, new investors with new relationships and fresh 5-7 (or 10) year horizons can step in and provide value they would have never had otherwise. Best case scenario is that original fund manager has seen enough success to take the position into a newer fund with a fresh timeline (assuming their LPs allow it). The liquidity allows them to restructure the opportunity to meet the needs of their investors.

Secondaries have changed the game for funds and founders alike because now everyone can take some chips off the table, sleep a little better and still be in it for the long haul. Investors should also recognize that a founder needing liquidity isn’t them giving up on their company, it’s making sure they can support themselves and their family and now have the peace of mind to focus on building a behemoth.

Understanding Secondaries

Secondaries are essentially private shares of a private company, and now there’s a market (or several) that allows you to buy into those companies, but it’s more complicated than one might think.

Take SpaceX for example. An early investor may have been in the company for 12-15 years, watching their position grow on paper with no way to realize any of it until secondaries. Now, with the company’s permission, they have the ability to sell off part (or all) of their position to a new investor and see some of the fruit of their patience. There are a lot of caveats here.

For one, companies don’t want people trading their shares on a regular basis before they’re public. Buying secondaries is like buying stock, but for smaller companies it also means buying control. That’s why larger firms like SpaceX have limits on how much people can sell or when so their ownership structure doesn’t suddenly change drastically. It’s also why some smaller companies don’t allow it at all. For almost every venture startup, if you want to buy shares directly you need the founder to be ok with the sale first and you second. Same goes for selling. Otherwise you’re stuck.

I went through this firsthand with Anduril. The company had a right of first refusal, meaning they get first dibs on taking the shares. They have an internal price limit for what they’ll buy at, and we were just over the threshold, something we didn’t know until after we closed. Even still, the process took months of checks with their legal department before I was cleared to buy. Today, you’d be hard pressed to find a means to get access to Anduril in direct secondaries because the company is particular about who directly owns their shares.

The Real Question

A quick gut check when looking at new opportunities, if you’re nervous about the timeline or size of the investment, you should probably rethink it.

How much of your capital can you afford to lock up for 5, 7, or 10+ years? Even if secondaries are an option how likely is that to be something you can or would pursue? What do you actually need, cash flow or growth? If this goes to zero, does it break your portfolio? And if everything goes perfectly but you can’t cash out when you need to, does that create problems elsewhere?

I’m not just saying this, I’ve had to practice what I preach. Not long ago I had to pass on a media investment opportunity, not because I didn’t think it was a good deal, but because it would have likely meant a long time horizon before seeing money and I wasn’t ready to make a commitment like that when I had so much else going on. Right now I’d probably be up multiples on paper, but I still wouldn’t be able to touch any of that money for years and I stand by my decision because of that. If you’re ever curious about the company, it started from Youtube and now does everything from TV to candy bars.

The best investors I know aren’t the ones who find the best deals. They’re the ones who understand which deals are right for them and have the discipline to pass on everything else. Every deal you take is a direct representation of how you treat money. Understanding someone’s liquidity needs, whether your own or the investor you’re pitching, is the difference between building something that compounds and getting stuck when you need flexibility most. Once you internalize that, everything else gets clearer.

Before evaluating any investment on its merits, ask yourself whether it actually fits your life and your timelines.

Disclosure: These are my opinions, not investment advice. I have positions in the sectors I write about. My views are shaped by my investments, and my investments are shaped by my views. Take everything with appropriate skepticism and do your own diligence.

Welcome to Investing in Organized Chaos, a platform where I’ll dive in to how I think about investing in all capacities. From real estate to stocks to venture and more. Sometimes I’ll be right, sometimes I won’t, but either way you’ll know what I’m thinking.

But first, some context on how I got here:

Advice I Ignored

When I first told anyone in my family or network that I wanted to pursue venture as a career full time the advice was consistent, go work for a fund, learn the ropes and then when you’re ready, go out and start your own thing.

I understood the rationale, but it felt too slow. It would take years to get to actually influence decision making, and even then I’d be limited to that fund’s style, their picks, their pipeline. I had this (perhaps irrational) confidence in my ability to succeed on my own.

To be fair it wasn’t completely irrational, like most Persian Jews, I was borderline raised to be a businessman. Thanks to my father, I started investing in the stock market when I was in middle school with the most basic advice “invest in companies and products you believe in that you think will be important in the world”. To say I loved it was an understatement, there was something gamified about watching the numbers go up on a screen. As I got older I learned what that truly meant and wanted to do it all faster, bigger, more precise. By the time I got to venture, I’d already spent years learning how to underwrite investments, understand people, mitigate risk, and trust my gut when it said to do something.

How I Actually Started

There’s a famous quote I love: “Luck is when preparation meets opportunity”. Whether I realized it or not, I had been setting myself up for my lucky break from the start. Through some families in my community, I got introduced to a venture fund investing in aerospace startups. What sounded like nonsense to my dad sounded like the Jetsons and Futurama to me, and I couldn’t wait to learn more. I read up on every company, learned about the sector, and ultimately decided I wanted in. At the time that was more or less 90% of every dollar I had to my name that I put into the fund, and as one does when investing that much of their net worth in anything, I became their most annoying minimum check ever.

I pushed to go on company tours, followed founders on twitter, subscribed to newsletters. At one point I started sending the fund updates on companies before they had them themselves.

Then came Anduril.

One of the fund’s portfolio companies got acquired by Anduril, and by some stroke of chance they were selling their position. It didn’t take long for me to match the highest existing bid and pursue the deal. Some would say luck, I would say I was prepared. I knew about the company, I spent years building a relationship with this fund, and I had just enough confidence to drop everything and go after it.

I won’t try to pretend like it wasn’t a big decision or wasn’t terrifying. I was walking away from a career I’d spent years building to chase an opportunity and a world I was just learning about. That’s when I got another piece of brilliant advice from my father “as long as you’re prepared to not make money for at least 1 to 2 years, then go for it”. It wasn’t pessimism about my ability to succeed, it was about understanding risk.

Paying it Forward

It took months, but we managed to get the deal done. Just when I was preparing to end my side quest and go back to real estate, the same investors came knocking and asking “that was great, what’s next?”

That was a great question. For me it was restarting the real estate fund I left to pursue Anduril, but for the fund I bought it from it was Impulse Space. I was excited about the company, willing to put the next big chunk of my net worth into it, and in an act of good faith for honoring the Anduril deal, sent investors that fund’s way too.

Paying it forward to them ended up being a circular action because they decided to cut me in on the vehicle as a Co-GP. To my surprise I ended up bringing a meaningful amount of investors to the opportunity, and not six months later I made my first money investing in venture from the fees.

That first check, that was the realization that this was more than a fun side quest for me, I wanted to do this full time. I loved learning about new exciting companies, bringing it to people I knew and getting equally excited about it. It didn’t feel like work, and I was good at it. I was good with connecting with people, building trust and relationships, and good at trusting my gut.

Suddenly people I had met on company tours, at conferences, everyone I spent years connecting with out of passion became my way in. I took a page out of my real estate playbook, there’s always a deal to be had, just find the structure.

I’d come up with creative structures to get my network access to companies that funds spent years building relationships with. This is where I broke the second piece of advice, have your name on the vehicle to establish yourself as a player in the game.

I thought, what’s more important, making the deal happen, or having my name on it? I chose the former, and in exchange got access to opportunities, people, and rooms I had no business being a part of, until I did.

Not Investing with “OPM”

A note on the Persian Jewish community in LA for those who are less familiar. This is a tight-knit group of people who fled religious persecution and established themselves from the ground up here in the US. Almost all of us did business with one another, maybe out of lack of trust for outsiders, maybe because of the language barrier, maybe because we were all just raised to see the process the same way. Whatever the reason, I was fortunate to have the first network of investors I could call on.

The difference for me starting out versus so many funds I’ve met is that I’m not investing with OPM (other people’s money), at least not really. My investors were my family, my friends, my neighbors. If someone had a problem or a question, they had my cell and they were not afraid to call me, whether it was 8am on a Sunday or 10pm on a Saturday.

In exchange for their trust, I treated their own money as well if not better than I did my own. If an investment were to go bad, that wouldn’t mean I’d have a harder time raising fund II, that would mean no fund II, it would feel like an injustice and disgrace to the people close to me who trusted me. Every deal I bring, I’m putting my own money in, and if not then I’m not sending it out. If we go down, we go down together, but I’ll be damned if I let us go down.

Translating Real Estate

It got to a point where the challenge wasn’t finding good opportunities to invest in, it was explaining to people that it was a good opportunity.

How do I get a community of people, traditionally invested in the most stable cash flowing asset class of all time in real estate, to something one rung shy of a casino on the risk scale with venture? It all came to translating.

Carry was promote, multiples are like cap rates, if people could understand the deal themselves, they could make educated decisions about whether or not they were up for the risk. You know how some people say it’s not rocket science? Well this time it literally was, and I was explaining it to a (albeit very educated) community where English was their second if not third language.

It took time, but slowly people’s trust in me helped them take small bits of risk, small checks here or there, but enough of them and I could piece together a real opportunity for us all. As time went on and companies grew, word spread, and quickly more and more people were curious about the kid from the community who was investing in rockets and drones.

I started building a track record, I earned more and more of people’s trust, and spent countless hours educating anyone and everyone on how it all worked until they felt comfortable. To this day I encourage anyone who wants to learn more to call me, I’ll find time for you. There’s nothing more satisfying than seeing that light switch on and they get as excited about this world as I am.

Realizing my Superpower

Every founder, be it a fund or a company, needs to know their superpower. I always thought I knew mine, but one day it dawned on me, I can connect with almost anyone. I’ve joked in the past that I could befriend a wall, but truly that’s how I started.

As I dove deeper into this new world I kept attending as many events as I could find and was not bashful about going up to the most influential or successful person there and waiting patiently to have any kind of conversation with them. A conversation with a Belgian aerospace CEO about where to get bratwurst in LA led to a private company tour with other funds and investors and later an opportunity to invest in even more opportunities.

I’ve had similar experiences with founders of all sorts of successful companies. The thing I reminded myself is that they’re just like anyone else, they have friends and interests and hobbies. They have stress with work and family and balancing everything, and at heart they’re more than happy to pass on the wisdom they’ve gained over the years if you’re willing to be respectful of their time and their words.

That mentality has gotten me introduced to countless founders, CEOs, investors, fund managers, and anyone I could ask to know, and I’m eternally grateful for each one of them sparing time to pass on their notes.

Being Open

Those relationships, those people taught me to ignore another piece of advice, it’s not only ok to share deal flow, it’s actually better.

I have nothing to hide, and that’s how I like to operate. I’m open with my opinions on companies for or against. I make introductions for funds, founders, and investors whenever I can, and spare my time and give advice when it is helpful.

In this world of hardware and innovation, it really takes a village. The more people who care about the success of a company, the more likely it is to succeed, so I’m as open as possible about my opinions and my access. Besides, reputation travels.

If someone tried to circumvent me on an investment, or a company cut me out, it speaks volumes on their character and probably saved me from being in a worse situation with them later. But now other funds, other companies, other investors probably took notice too, and people don’t forget easily.

You’re Never Doing it Alone

Even with all these things moving positively, having built connections, deal flow, shown returns, it’s not exactly a smooth process, especially going at it alone.

I chose to be a solo fund manager because I have an unreasonably high expectation for myself and my work ethic that wouldn’t be justified to put onto a partner or employee. I was willing to own that every hour I worked meant progress I benefitted from and that if things went south there was no one to blame but me.

That remains true for fundraising, it’s a constant effort to put yourself out there and more often than not you get turned down. The emotional toll is real, and I can’t count how many times I almost fully broke down and wanted to quit. Over and over I thought, why don’t I just go back to real estate?

Those thoughts can consume you, and for me they almost did more than once, but thankfully I’m supported by so many amazing people in my life. People who know me, trust me, and believe in me, and remind me that if they didn’t think I could do it, they never would have supported it.

Where I Am Now

After closing Anduril in June of 2024 I chose to make this my life. Nearly 15 investments, a new fund, dozens of conferences, a lot of ignored advice and a few breakdowns later, I’m still standing.

What You Can Expect

The idea of this project is to talk all things investing and hopefully provoke some people to look into new opportunities, and remember that there’s more than one right path to success.

Take every call, go to every event, don’t be afraid to start conversations. Even the most successful people in this world are still just people. Be respectful of their time, find something real to connect on, and see what happens.

Welcome to Organized Chaos.

Disclosure: These are my opinions, not investment advice. I have positions in the sectors I write about. My views are shaped by my investments, and my investments are shaped by my views. Take everything with appropriate skepticism and do your own diligence.