The Great Reinvention logo THE GREAT
REINVENTION

Investing in the Frontier: A National Innovation Fund for Fusion, Longevity, AI, and Space

Proposed legislation: The Frontier Sciences and National Innovation Act

Innovation Funding: A Cost-Benefit and Investment Analysis

The United States built the modern world on a bet that public investment in fundamental science pays for itself many times over. The transistor, the internet, GPS, the polio vaccine, the Human Genome Project, and the GPS chip in every phone all trace back to federally funded research that no private firm would have financed on its own. Yet today the federal government's commitment to that bet has quietly eroded. According to the National Science Foundation's National Center for Science and Engineering Statistics, federal funding for research and development accounted for roughly 0.63 percent of GDP in 2024 — down from a Cold War peak of 1.86 percent in 1964. The country that invented the future is now spending a third of what it once did, as a share of its economy, to keep inventing it.

This page proposes a dedicated Frontier Sciences and National Innovation Fund of $30–50 billion per year — a sustained, mission-driven commitment to four domains where breakthroughs are both scientifically ripe and strategically decisive: fusion energy, life-extending medicine, artificial intelligence for the public good, and space exploration. The argument is not that government should pick commercial winners. It is that the federal government should do what only it can do: fund the high-risk, long-horizon, capital-intensive science that markets systematically underfund, and that adversaries are now funding aggressively.

To put $30–50 billion in context: the entire National Science Foundation received about $7.5 billion in R&D obligations in FY2024, and the National Institutes of Health roughly $52.5 billion, per NSF's federal funds survey. A Frontier Fund at this scale would not replace those agencies — it would add a focused, mission-oriented layer on top, comparable in ambition to the Apollo program or the early years of DARPA.

What Gets Built and Funded

The Frontier Fund is organized around four mission directorates, each with concrete deliverables rather than open-ended grants.

Fusion energy. Fusion has crossed a genuine scientific threshold. On December 5, 2022, the National Ignition Facility at Lawrence Livermore National Laboratory achieved the first controlled fusion reaction in history to produce more energy than the laser energy delivered to the fuel — 3.15 megajoules out from 2.05 megajoules of laser input, per the Department of Energy. Ignition has since been repeated multiple times, with a February 2024 shot producing roughly 5.2 MJ. The DOE has begun building out the field with programs such as the $42 million inertial fusion energy hubs and the $180 million FIRE Collaboratives. But these figures are modest relative to the prize. The Frontier Fund would dedicate roughly $8–12 billion per year to a portfolio of approaches — inertial confinement, magnetic confinement (tokamaks and stellarators), and private milestone-based contracts — with the explicit goal of a grid-connected pilot plant in the 2030s.

Life-extending medicine. The National Institute on Aging received about $4.5 billion in FY2024, per NIA's own budget documents — a small slice of NIH and a tiny fraction of what the U.S. spends treating the diseases of aging after they appear. The Frontier Fund would direct roughly $8–12 billion per year toward the biology of aging itself: cellular senescence, metabolic regulation, regenerative medicine, and large longitudinal trials. The logic is "geroscience" — the hypothesis that targeting aging mechanisms could simultaneously delay multiple age-related diseases. Even a modest compression of late-life morbidity would represent enormous economic and human value.

Artificial intelligence for the public good. AI capability is concentrating in a handful of private firms with the capital to build billion-dollar training clusters. The National AI Research Resource (NAIRR), led by NSF in partnership with 13 federal agencies and dozens of private contributors, launched as a pilot in January 2024 and has already supported roughly 600 research projects, per NSF — but it remains a pilot stitched together from donated compute. The Frontier Fund would put roughly $8–12 billion per year into a permanent public AI infrastructure: sovereign compute for universities and national labs, open scientific foundation models, evaluation and safety research, and the DOE's emerging effort to connect national-lab supercomputers and datasets into a unified scientific AI platform.

Space exploration. NASA received about $24.9 billion for FY2024, with roughly $8.1 billion tied to the Moon-to-Mars exploration architecture, per NASA budget documents and the Planetary Society. The Frontier Fund would add roughly $6–14 billion per year — not to duplicate NASA, but to accelerate the science-and-capability layer: in-space manufacturing, nuclear thermal and electric propulsion, lunar resource utilization, and next-generation space telescopes that extend the scientific yield that produced JWST and Hubble.

Cost Breakdown

Mission directorate Annual range
Fusion energy $8–12B
Life-extending medicine $8–12B
AI for the public good $8–12B
Space exploration $6–14B
Total $30–50B

At the high end, $50 billion is roughly 0.7 percent of the FY2024 federal budget and well under one-fifth of one percent of GDP. It would roughly double total federal R&D as a share of GDP back toward late-1990s levels, still far below the Apollo-era peak. Crucially, the fund is designed to be counter-cyclical and durable: a ten-year authorization with milestone-gated tranches, so that programs are judged on results rather than annual appropriations fights.

Benefits

Economic returns. The economics literature on public R&D is unusually consistent: social returns substantially exceed private returns, which is precisely why markets underinvest. Estimates of the social rate of return on public research commonly run well into the double digits annually — figures cited across work from the Congressional Budget Office, Brookings, and academic economists. The federal investments behind the internet, GPS, and the Human Genome Project are routinely estimated to have returned many times their cost. While any single estimate should be treated cautiously, the direction is not in dispute: sustained basic-research funding is among the highest-return uses of public money available.

Jobs and industrial base. Frontier research anchors high-wage employment — physicists, engineers, technicians, and the supply chains around national labs and university centers. A fusion supply chain alone spans superconducting magnets, precision optics, advanced materials, and power electronics, much of which overlaps with semiconductor and defense manufacturing. The aging-biology and AI directorates would expand domestic biotech and compute capacity at a moment when both are strategically contested.

Security and strategic position. The competitive context is stark. Private fusion investment globally has reached nearly $10 billion over five years, per reporting compiled by The Conversation and the American Nuclear Society, with U.S. firms like Commonwealth Fusion Systems (roughly $3 billion raised) and TAE Technologies (roughly $1.2 billion) leading commercially. But on government deployment spending, China is reported to have spent at least $6.5 billion on fusion between 2023 and 2025, per CNBC and SCSP analysis — more than triple comparable U.S. public spending. The same dynamic is visible in AI compute and quantum. Frontier leadership is no longer a luxury; it is the substrate of economic and military power for the rest of the century.

Administrative and Implementation Considerations

A fund this large fails if it becomes a slush fund. Three design principles matter.

First, mission directorates with measurable goals, on the DARPA and Operation Warp Speed model: empowered program managers, milestone-based contracts, and the authority to terminate underperforming efforts. DARPA's structure — small teams, finite project terms, tolerance for failure — is the template, not the sprawling formula-grant model.

Second, milestone-gated tranches. Rather than appropriating $50 billion annually with no strings, the fund releases capital as programs hit independently verified technical milestones (e.g., a fusion plasma sustaining a defined gain, a longevity biomarker validated in trial). This protects taxpayers and concentrates money where progress is real.

Third, coordination, not duplication. The fund should route money through existing institutions — DOE national labs, NIH/NIA, NSF, NASA, and qualified private contractors — under a small coordinating office (modeled on the National Quantum Coordination Office) rather than building a vast new bureaucracy. A standing independent review board, drawing on the National Academies, would publish annual progress assessments.

International Comparisons and Precedent

The clearest precedent is American: the Apollo program at its peak consumed over four percent of the federal budget and produced not just a moon landing but integrated circuits, software engineering, and a generation of scientists. The post-war NIH and NSF buildout, and DARPA's seeding of the internet, are the same story.

Internationally, the ITER fusion megaproject — a collaboration of the EU, U.S., China, Russia, Japan, Korea, and India — illustrates both the promise and peril of frontier science: its cost has grown to well over $25 billion, per the Congressional Research Service, with repeated schedule slips, even as it advances the physics. China's national fusion buildout, its quantum program, and the EU's Horizon Europe research framework all reflect a global recognition that frontier R&D is a contest. The lesson from ITER is to favor a diversified portfolio with multiple competing approaches over a single monolithic megaproject.

Comparison to the Status Quo and Alternatives

The status quo is not "no spending" — it is fragmented, under-scaled, and annually contested spending. Fusion gets tens of millions where it needs billions; aging biology gets a sliver of NIH; public AI compute is a donated-hardware pilot; space science competes against operational programs inside a flat NASA topline. Each is individually defensible and collectively inadequate to the moment.

One alternative is to rely on private capital, which is genuinely surging in fusion and AI. But private markets discount long horizons and underprovide the open, non-excludable knowledge that fuels everyone's progress. Private fusion firms still depend on DOE facilities and decades of publicly funded plasma physics. A second alternative is to simply enlarge existing agency budgets — defensible, but agencies optimized for steady-state grant-making are not structured for moonshots. The Frontier Fund's value is in adding a mission-driven, milestone-gated layer designed specifically for high-risk frontier work.

Risks, Trade-offs, and Counterarguments

The strongest objection is that government is bad at picking technologies, and that fusion in particular has been "thirty years away" for fifty years. This is a fair critique. The honest response is twofold: first, the science has genuinely changed — repeated net-energy ignition at NIF is not a press release but a measured result; second, the fund is deliberately technology-agnostic within each mission, spreading bets across approaches and killing losers via milestone gates rather than committing to one champion.

A second objection is opportunity cost: $30–50 billion a year is real money that could fund tax cuts, deficit reduction, or social programs. This is true, and the case rests on returns: if public R&D earns even a fraction of its historically estimated social returns, the Frontier Fund is among the highest-yield uses of the marginal federal dollar. But returns are probabilistic and back-loaded, which is precisely why they are politically hard to defend and why durable, multi-year authorization matters.

A third concern is hype and misallocation — longevity science attracting charlatans, AI funding chasing fashionable applications, fusion timelines oversold. Independent technical review, transparent milestones, and a willingness to publish failures are the only real safeguards. There is no version of frontier funding without risk; the policy choice is whether to manage that risk or cede the frontier.

Finally, there is the distributional question: do the benefits of life extension, AI, and advanced energy accrue broadly or to the already-advantaged? Designing the AI and longevity directorates explicitly around public access — open models, open data, and broad clinical participation — is essential to the fund's legitimacy and is built into the proposal's framing as innovation "for the public good."

Conclusion

The United States did not become a scientific and economic superpower by accident. It made a sustained public bet on fundamental research, and that bet paid off in industries, medicines, and capabilities that defined the age. That bet has quietly shrunk to a third of its Cold War scale even as the strategic stakes have risen and the science — in fusion, in aging biology, in AI, in space — has never been riper. A $30–50 billion annual Frontier Sciences and National Innovation Fund, structured around mission directorates with hard milestones and independent review, would restore the country's commitment to its own future at a cost well under one-fifth of one percent of GDP. The risk is not that the money is wasted; the risk, given who else is now spending, is that we do not place the bet at all.

Sources

← Back to The Great Reinvention