After 14 years in the aerospace industry, I recently moved to fusion. I was drawn to fusion for the same reason I was drawn to aerospace: an intergenerational goal that demands technical rigor and complex coordination across disciplines and systems, with huge potential upsides for humanity.
The two industries share more structural features than is often appreciated. Both pursue capabilities once considered the exclusive domain of national governments. Both require the integration of advanced materials, cryogenics, high-power electronics, and precision manufacturing at scales that take decades to mature. Both depend on long-horizon capital that sits uneasily within standard venture timelines. And both have followed a similar institutional progression — from state-led megaprojects driven by geopolitical competition, to international cooperation efforts, to a more recent wave of privately funded companies pursuing diverse technical approaches.
Decisions being made today about fusion regulation, public-private partnership structure, capital allocation, and inter-company collaboration are being justified in part by reference to what worked or failed in the space industry. That makes it worth examining carefully. This post addresses three questions:
- Which structural features of the space industry’s evolution actually translate to fusion, and which break down on closer inspection?
- What lessons from the space industry’s transition to private commercialization are most directly applicable to fusion’s current stage?
- Where will fusion need to develop its own playbook — because the underlying problem differs from space in ways the analogy obscures?
The answers point to a coherent set of policy choices, drawn from space’s hard-won lessons, that could position U.S. fusion to lead. Some of the groundwork has been laid, but the current federal commitment is underfunded and fragmented.
Parallel histories
The US space launch industry went through roughly three eras. The Apollo Program was a single architecture with state-driven, geopolitical motivation. This era of intense competition, called “the space race”, offered no promise of economic return – just the prestige and validation that came with beating the USSR to the moon. This era was succeeded by the development of the Space Shuttle and ISS, which brought international cooperation, megaproject overruns, and design-by-committee critiques. With the Space Race won, the absence of intense competition led to some slowdown in both federal funding and the pace of innovation. Then in the early 2010’s, the New Space era introduced commercial demand, venture capital, and fresh innovation. New technocratic rivalries emerged, such as SpaceX vs. Blue Origin or, in the smallsat arena, RocketLab vs. Stoke Space.
Fusion has followed a similar arc to the space launch industry. Both rocketry and fusion research projects were initially funded for military applications (see: ICBM’s and H-bombs). The early fusion era was led by national labs in the US, UK, and Russia, with machines like the T-3 Tokamak and the ZETA Z-pinch. ITER is the international megaproject of fusion’s middle era– starting in 1988 and projected to begin operations in 2035, it will cost at least $22B between 35 member countries. Like the middle era of the space industry, fusion innovation stagnated somewhat in the absence of competition and geopolitical urgency. The current private fusion era was galvanized by technological advancements in high-temperature superconductors (HTS), power electronics, and AI, plus the growing demand for non-fossil power generation.
In the case of the space launch industry, significant cost reductions weren’t achieved until its third era, approximately 50 years after the Space Race began. What conditions led to this inflection point, and which of them are reproducible for fusion? Answering this is the underlying work of 1cFE, which uses calibrated technoeconomic models to map feasible corridors to ultra low-cost fusion power across a range of future scenarios. In the sections below, we break down those enabling conditions and identify which are most applicable to accelerating commercial fusion.
Where the analogies break— Fusion vs. Space
The players are different. While the Cold War space race was driven by the United States and Russia, today’s race to commercialize fusion power is largely led by the United States and China. The Cold War space race was symmetric: two state programs running essentially the same architecture (multi-stage liquid propellant rockets) toward the same goal. The US-China fusion competition is largely asymmetric. China runs an Apollo-style program centered around a single architecture (D-T tokamaks) with strong government support, a large trained workforce, and a rapidly-maturing supply chain. The U.S. runs a portfolio model: dozens of private companies pursuing different physics, supported by private investors and, to a lesser extent, federal mechanisms like ARPA-E and the DOE Milestone Program. But the interesting question isn’t who produces net power first— it’s which model is better suited to converge on a commercially-viable power plant given the current state of unresolved physics and engineering.
Fusion has a higher technical bar than space did at commercialization. Aerospace had working demonstrations to build from before private capital arrived. NASA demonstrated orbital flight and lunar landing in the 1960s. While most of NASA’s heritage designs are classified under International Traffic and Arms Regulations (ITAR), private companies like SpaceX and Blue Origin were given reasonably good blueprints to work from. In contrast, fusion has not yet demonstrated net-power generation. NIF achieved scientific breakeven (Q_plasma > 1) in December 2022, but engineering breakeven — Q_engineering > 1, accounting for end-to-end plant efficiencies — has not yet been demonstrated. Private fusion companies are betting on physics breakthroughs and engineering cost reductions simultaneously, which is a structurally riskier proposition than what New Space took on.
Despite this, capital keeps flowing. As seen in the graph below, fusion equity investment in 2025 surpassed the historic 2021 peak to set a new annual fundraising record. Per the Fusion Industry Association’s 2025 Global Fusion Industry Report, cumulative industry funding now stands at $9.7B across 53 surveyed companies. China, meanwhile, has spent at least $6.5B on public and private fusion projects since 2023, focusing mainly on D-T tokamak architectures.
The customer problem is fundamentally different. Launch demand has consistently exceeded supply for over a decade. Every new rocket company has had a built-in customer base across NASA, DoD, ESA, and the increasing number of private satellite constellations. Conversely, fusion power will enter a competitive power market with many low-cost alternatives that are continuously improving. Forward-thinking tech companies such as Google and Microsoft have signed Power Purchase Agreements (PPA’s) with CFS and Helion, respectively – but these symbolic First-Of-A-Kind agreements will only persist if fusion power plants are able to produce electricity at competitive rates.
Even so, multiple New Space companies have pivoted from launch provision to higher-margin or higher-demand industries after disappointing ROI’s. Faced with SpaceX dominance and cooling VC interest, multiple New Space companies have had to pivot from their original launch theses in order to survive. SpinLaunch has redirected towards developing a comm-sat constellation, abandoning kinetic launch as its near-term revenue path. Astra leaned into in-space propulsion via its acquired Apollo Fusion business, reducing its near-term reliance on Rocket 4. ABL and Stratolaunch exited commercial launch entirely, repositioning around missile defense (ABL) and hypersonic flight testing (Stratolaunch) for the DoD.
We are already seeing early signs of similar hedging in the fusion industry, even before any company has faced the aforementioned “customer problem”. There should be more room for profitable players in the fusion industry given its potential market size, but electricity is a pure commodity with little room to differentiate on anything but price. And fusion’s competitors — gas, solar, fission, batteries — are continuously adapting and improving. Despite these structural differences, fusion companies can still draw clear lessons from the New Space survivors, which will be explored in detail in the following sections.
Where the analogies hold – Fusion vs. Space
Public-to-portfolio transition. Both space and fusion have moved from government-led megaprojects toward more privately funded, diverse approaches — but the transition looks very different in the US and China. China’s significant government funding remains focused primarily on a single architecture: the D-T tokamak. This is a defensible bet given China’s record-breaking progress in that domain. The US, by contrast, is funding dozens of reactor concepts — tokamaks, stellarators, magnetic mirrors, magneto-inertial, laser inertial, aneutronic fuels — through a mix of private capital and, to a lesser extent, federal programs. Whether this portfolio diversity proves to be inefficient diffusion or prescient hedging won’t be clear for some time. But the space analogy offers some reassurance for the U.S. fusion industry: had the New Space era focused only on the architecture requiring the least technical development, the breakthroughs in additive manufacturing, advanced propulsion, and reusability might never have materialized.
A step-change unlock. In the space sector, launcher reusability was perhaps the greatest insight of the New Space wave. A reusable booster (and, in some cases, a reusable second stage) allows for increased launch cadence and lower cost per kg to orbit. Fusion has no consensus equivalent — different companies are betting on different unlocks. HTS magnets enable smaller, higher-field tokamaks. Aneutronic fuels like D-³He and p-B11 and would reduce costs associated with neutron damage and radioactivity. Like reusable rockets, these potentially transformative technologies will require significant upfront investment before any long-term benefits are seen.
Cost has historically been treated as secondary to performance in both fields. Apollo, the Shuttle, ITER, and JWST all optimized for capability rather than unit economics. New Space changed that for rockets. SpaceX drove launch costs down through reusability and vertical integration, and companies like LEAP 71 are now employing AI optimization to drastically accelerate design timelines. Fusion is starting to follow the same playbook in different ways. Commonwealth Fusion Systems is manufacturing its own HTS tape at scale, while DeepMind has demonstrated reinforcement-learning control of plasma instabilities in tokamaks. Thea Energy is exploring modular, planar coils on its Helios reactor, which should improve the manufacturability issues with stellarators. These companies have prioritized manufacturability while exploiting recent technological developments across industries.
Transferable lessons from space to fusion
Risk-based regulation matters. This seems like an obvious statement, but improper regulation could end the fusion industry before it begins. To illustrate this, let’s consider the commercial drone industry. Pre-2016, U.S. drone operators had to individually petition the FAA for case-by-case exemptions designed for much larger manned aircraft — a process that typically took six months or more. Only after the FAA introduced a risk-based regulatory framework (Part 107) did the commercial drone industry scale rapidly, now valued globally at ~$15B and rising.
The fusion equivalent is the NRC’s October 2023 decision to regulate fusion under 10 CFR Part 30 (byproduct material) rather than Part 50 (the utilization facility framework used for fission). Applying Part 50 to fusion would have been like applying Boeing 737 rules to a quadcopter — a framework built for a different risk profile, and compliance costs that would likely have killed most private fusion economics.
The next regulatory question is whether the framework will further differentiate between neutronic and primarily aneutronic approaches. The radiological risk profiles are meaningfully different, and a one-size-fits-all approach within Part 30 leaves performance on the table. Primarily aneutronic fuel companies such as Helion and TAE are thus lobbying the NRC for risk-based regulations that recognize their significantly lower amounts of radioactivity as compared with a D-T fusion plant. Innovation stalls when regulation is poorly fitted to the technology it governs.
Megaproject opportunity cost is the real cost. ITER, JWST, and SLS share a critique that goes beyond schedule and budget overruns: they crowd out the rest of the portfolio. JWST cost ~$10 billion and forced a $1.4 billion reallocation from other NASA astrophysics missions. Astronomer Adam Frank has famously asked whether the field could have produced more science with ten $1 billion missions instead of one $10 billion observatory. ITER absorbed a generation of public fusion funding that could have supported a more diverse portfolio of smaller experiments. These megaprojects cost not only their budget (often inflated due to bureaucratic impediments) but also the optionality of everything that didn’t get funded.
The New Space response to this was incremental, fail-fast development — SpaceX’s “fail early and often” approach to Falcon 9 development, where rapid iteration on hardware that occasionally explodes turns out to be cheaper and faster than the zero-failure ULA approach. Fusion is starting to adopt the same posture. CFS will use the SPARC (Smallest Possible ARC) demonstrator to validate magnet and plasma performance well before the full ARC reactor gets built. OpenStar Fusion is developing its second of four reactor iterations on the way to a commercial powerplant. Helion has built and tested seven incremental prototypes to date. Helion recently unveiled its “Tiny Merge”, a new testbed less than one-eighth the size of its latest machine Polaris, designed to run experiments on FRC formation and merging. With a 2028 Microsoft delivery deadline rapidly approaching and fundamental physics questions still open on Polaris, Helion essentially concluded that the seventh-generation machine alone wasn’t iterating fast enough. The lesson is that “fail-fast” doesn’t mean abandoning incremental scaling, but running the iteration loop on the cheapest hardware that can answer the next question.
Not every fusion company is taking the incremental approach. Pacific Fusion emerged from stealth in 2024 with a $900M Series A and plans to go from a clean sheet design to a demonstration system of approximately 156 identical 2TW pulser modules, targeting net facility gain by 2030. There is planned iteration and validation at the pulser component level, but no intermediate integrated fusion machine between the component tests and the full demonstrator system. The bet is that capital plus speed beats iteration — a defensible position given fusion’s compressed timeline against Chinese competition, but the opposite of the SpaceX-style incremental development that defined New Space’s most successful players.
Public-private partnerships work when the government acts as a customer, not a designer. NASA’s COTS (2006) and Commercial Crew (2014) programs are the archetype. NASA defined what it needed (cargo or crew transport to the ISS), set fixed-price milestones, awarded multiple companies to preserve competition, and let the contractors design the vehicles. It was the COTS program that helped to save SpaceX from bankruptcy after multiple failures of their Falcon 1 rocket. The result: SpaceX delivered Crew Dragon at roughly half the cost of Boeing Starliner and on a faster timeline. The closest fusion analog is the DOE’s Milestone-Based Fusion Development Program, launched in 2023 with $46M to eight companies. The structure is effective, but the funding level is not yet at the scale needed to materially accelerate first-of-a-kind power plants.
Consider Alternate Revenue Streams. The only consistently profitable private launch provider is SpaceX, and that is almost entirely because of Starlink. Falcon 9’s launch business alone barely breaks even at 75% gross margins; SpaceX is a profitable company because it anticipated a downstream, recurring-revenue product that its launch capability uniquely enabled. The lesson for fusion is that betting solely on electricity sales (a low-margin commodity) may not produce a profitable fusion company even if the technology works. Successful fusion firms will need to anticipate and capture downstream value streams. Marathon Fusion’s idea to use fast neutrons from the fusion reaction to transmute mercury to gold is one such revenue stream. Other candidates include radioisotope production, process heat-as-a-service, and direct sales of HTS tape or pulsed-power electronics to the medical or defense industries.
Co-design with the customer in mind. Cowboy Space (formerly Aetherflux) has recently raised $275M to build its own rockets after concluding that no commercial launch provider could scale its orbital data center business fast enough. Cowboy Space isn’t building a generic rocket and selling it to data center customers — they’re building a rocket where the data center is the second stage. Co-designing payload and vehicle as one product eliminates entire integration layers and the duplicated systems they require. The fusion equivalent is co-designing the plant and the customer’s facility, perhaps a large data center, from the start. A co-located fusion plant can share cooling infrastructure, security, and grid interconnection costs with an AI data center. Some fusion fuels and reactor architectures can also output DC power directly and modulate output to match real-time demand, eliminating AC-DC conversion losses and reducing the on-site storage that data centers typically require. For fusion plants using steam turbines, the large quantities of waste heat — typically 50–60% of total thermal output — could be routed through absorption chillers to provide data center cooling, thus reducing one of the data center’s largest operating expenses. Sharing infrastructure with an external partner brings real complications around ownership, liability, and design control. But the CAPEX and operational synergies are large enough that the fusion companies willing to navigate this complexity may end up with a meaningful cost advantage over those who design in isolation.
Cross-industry collaboration is more important for fusion than it ever was for space. Though many aerospace companies share launch infrastructure, test stands, and strategic vendors, direct cost or IP sharing between competing space companies is essentially nonexistent. Fusion companies may need to chart a new course here, given their structural incentive to collaborate: a large number of relatively small fusion companies are trying to solve many shared technical problems, from plasma control to fast neutron bombardment to tritium breeding to cycle-limited power electronics. It’s unlikely that any single company will solve all of these issues better than the rest. And unlike the space sector, no individual fusion company can credibly claim to reach the revenue stage within a typical VC’s timeline.
This is why the fusion industry needs more collaboration and support than space did if the U.S. wants to continue as a leader in fusion power. The Special Competitive Studies Project (SCSP) has suggested a $10B federal fusion investment to ensure that the US beats China to market. Roughly 50% of this funding would go to public R&D investments to address many common knowledge gaps across the industry. 40% would go towards building demonstrator plants for a select few fusion companies. 10% would go into public-private mechanisms such as the Milestone-based development program and research grants. Whether this is the right structure for the money is a question worth examining carefully, which I will return to at the end of this post.
In the private sector, the CFS-Realta partnership announced earlier this year is a notable example of symbiotic collaboration. CFS is supplying integrated HTS magnet systems for Realta’s prototype and eventual power plants — a deal CFS describes as having multi-billion-dollar potential. Two companies pursuing different fusion architectures (tokamak and magnetic mirror) are sharing manufacturing capability and supply chain access. Similarly, Avalanche Energy’s shared lab space called FusionWERX has potential to collectively benefit small fusion companies while generating intermediate revenue streams for Avalanche.
What this means for fusion’s next few years
New Space peaked around 2021–22, and the shakeout has been ongoing. The companies that survived built cost discipline, customer-driven design, manufacturability focus, rapid iteration, and incremental testing into their cultures from the start. They also have successful public-private partnerships and innovative alternate revenue streams (see: Starlink).
The transferable lessons from space to fusion are important: risk-based regulation enables industries while misapplied regulation kills them, a rapid, incremental approach beats out the megaproject approach, public-private partnerships work when governments buy outcomes rather than specific designs, customer-driven design can unlock economic synergies, and alternate revenue streams are essential in high-risk industries.
But the analogy breaks where it matters most. Fusion has not yet demonstrated engineering breakeven, and its customers — utilities buying a pure commodity in a market full of cheap, improving alternatives — bear no resemblance to the aerospace customers with limited launch options. Fusion companies are betting on physics breakthroughs and cost reductions simultaneously, which is structurally riskier than what companies like SpaceX faced.
China and the U.S. are facing these commercialization risks in very different ways, as discussed previously. China is relying on its large government funding, skilled workforce, and built-out supply chain to commercialize the D-T Tokamak. To counter this, the SCSP has proposed a $10B U.S. investment across three categories: public R&D infrastructure to close scientific gaps, public-private partnerships with milestone and grant mechanisms, and a demonstration tier to help fund first-of-a-kind powerplants. In an October 2025 report, the SCSP stated “The race [between the US and China fusion industries] is no longer theoretical; it is unfolding now, and the consequences of losing would reverberate across energy security, economic leadership, and national power.“
While I agree with many of the recommendations from the SCSP, this framing seems a bit oversimplistic, as if trying to catalyze fusion innovation in the same way that Kennedy and Khrushchev catalyzed the space race. Being first to market is not the same as winning the market. Take SpaceShipOne, which won the Ansari X Prize in 2004 and proved private human spaceflight was possible. Its successor, Virgin Galactic, has been unable to successfully monetize this technology because its suborbital architecture fundamentally limits its market reach. This is a prime example of designing for a milestone rather than a market demand. Fusion’s frontrunner architectures could face a similar fate if their complexity, scale, or operating costs preclude them from competing in future power markets.
How smart fusion policy could build the next SpaceX
Concentrating federal support on a small number of bets early in a technology’s development can foreclose the diversity that often produces the winners. The U.S. fusion industry will require sustained federal support to remain a global leader, and the SCSP is right that significant funding is warranted for public R&D infrastructure. The SCSP’s proposed $4B “demonstration tier”, however, deserves closer scrutiny. Currently, the SCSP’s demo tier is budgeted around two DOE-chosen FOAK plants at roughly $2B each. This structure is modeled after fission’s ARDP, which has shown somewhat dubious results to date. If this demo tier were restructured to mirror NASA’s COTS program, it may be possible to support a wider range of technologies without overextension. Fixed-price milestone payments would cap the DOE’s exposure, and multiple parallel awards with clear reallocation when companies miss milestones would incentivize timely results. Furthermore, DOE Loan Programs Office credit support — direct loans or loan guarantees — can stretch federal dollars further by financing construction at low interest rates, with (initially) subsidized revenue providing the bankable cash flows to service the debt. In this way, the same federal commitment that SCSP would spend on two grants could possibly support construction financing for 4-6 companies, without picking technology winners or absorbing cost overruns.
Revenue-side support has to extend past first-of-a-kind. A COTS-style demo tier gets the first fusion plants built. It does not, on its own, drive the kind of cost reductions that turn a technology into a competitive commodity. That second phase has historically required sustained, technology-specific revenue support at scale. The two clearest precedents are Germany’s Renewable Energy Sources Act (EEG), which paid out roughly €100 billion in cumulative feed-in tariffs for solar through 2020, and the UK’s Contracts for Difference (CfD) program, which has paid out ~£8.9 billion since 2014 and contracted 39 GW of renewable capacity. Both programs absorbed real cost from consumers and taxpayers in exchange for a learning curve that bent steeply downward.
Fusion will learn slower than solar did. Wright’s-law learning rates run roughly 20–30% per doubling of cumulative capacity for solar PV, ~35% for lithium-ion batteries, and ~10% for wind. Fission, by contrast, exhibits negative learning across most national programs, due to its large, bespoke nature and high regulation. A realistic planning assumption is that fusion’s learning rate will lie somewhere between fission and solar.
That argues for a federal CfD or strike-price program sized to support roughly 4-8 GW of cumulative fusion capacity over ten years, at a cost of $15–30 billion in support payments. This is meaningfully larger than the SCSP’s $10B proposal, but an order of magnitude smaller than what Germany spent seeding the global solar industry. It is sized to fund 3–4 doublings of installed fusion capacity, which is the minimum needed to test whether fusion’s actual learning rate is closer to fission’s or to renewables’. This need not come at the expense of other clean-firm technologies. SMRs, geothermal, and long-duration storage each have a similar case for revenue-side support, and a combined clean-firm program would still be a small fraction of the ~$800 billion the CBO projects for federal clean-energy tax credits over the next decade. A subsidized revenue stream of this scale would kickstart the fusion industry on a technology-agnostic basis, rewarding companies that deliver concrete results — and giving the U.S. the empirical basis to decide which architectures deserve scaling further.
In closing, the Space Race catalyzed innovation for both the U.S. and Russia, but it wasn’t until the New Space era that a competitive launch market emerged. If fusion were to follow that blueprint, it could be decades from first net power to a commercially viable plant. With these lessons in mind, fusion policymakers and founders should prioritize fusion technologies not only for their near-term technical feasibility, but for their commercialization potential in future energy markets.