AI, Fusion and Batteries: Deep Tech’s Billion-Dollar Cycle
Deep tech is having a new “show me the physics” moment. After two years in which venture investors obsessed over software efficiency, burn multiples and the fastest path to revenue, some of the largest recent startup checks are flowing into companies that may not look like conventional venture bets at all: a frontier AI lab with no consumer app, a fusion company racing to turn a tokamak into a power plant, and a thermal-battery startup trying to electrify the hottest corners of heavy industry.
That is the signal inside Crunchbase’s latest funding roundup: capital is clustering around Safe Superintelligence, Commonwealth Fusion Systems and Antora Energy. Different markets, different timelines, same investor logic. The next cycle of tech value may be defined less by another SaaS workflow and more by who controls intelligence, energy abundance and industrial electrification.
The New Deep-Tech Barbell
The striking part of this funding pattern is not simply the size of the rounds. It is the shape of the risk.
On one side is AI: capital-light in theory, but now brutally capital-intensive in practice because compute, talent and model training costs have become the moat. Safe Superintelligence, founded by OpenAI co-founder Ilya Sutskever with Daniel Gross and Daniel Levy, has been explicit about its mission: build safe superintelligence, full stop. That narrowness is unusual in an AI market crowded with copilots, agents and enterprise wrappers.
On the other side are fusion and batteries: physically complex, manufacturing-heavy and dependent on years of engineering execution. Commonwealth Fusion Systems is trying to commercialize fusion power using high-temperature superconducting magnets. Antora is pursuing a nearer-term industrial decarbonization play, storing renewable electricity as heat in solid carbon blocks and delivering that heat back to factories.
The barbell is clear: investors are underwriting both the most abstract form of leverage, intelligence, and the most concrete constraint on growth, energy.
Safe Superintelligence Shows AI Is Still the Capital Magnet
Safe Superintelligence represents a notable shift in AI startup formation. It is not selling a chatbot for legal teams or a coding assistant for enterprises. It is a research company built around a single thesis: the race toward superintelligent systems is real enough, and risky enough, to justify a purpose-built lab.
That would have sounded academic a few years ago. Now it is financeable at extraordinary scale. Reuters reported in 2024 that SSI raised $1 billion only months after launch, backed by investors including Andreessen Horowitz, Sequoia Capital and DST Global. The point is not merely that elite investors want exposure to the next OpenAI. It is that the AI funding market has matured into a fight for scarce inputs: world-class researchers, training infrastructure, proprietary technical approaches and time.
The biggest AI companies are already spending like infrastructure players. Frontier-model labs need clusters, chips, energy contracts, safety teams and long experimentation cycles. That makes AI look less like classic software and more like a strategic industry. SSI’s appeal is that it offers a pure-play position on the highest-upside version of that industry, without the distraction of near-term enterprise packaging.
The risk is obvious: no product, uncertain timelines and a technical goal that may prove elusive. But in this cycle, that kind of risk can be acceptable if the prize is foundational. Investors are not just buying revenue growth; they are buying optionality on the control layer of the future economy.
Commonwealth Fusion Makes Energy Abundance Investable
If AI is creating demand for power, fusion is one of the grand bets on supplying it. Commonwealth Fusion Systems has become the flagship private fusion company because it has paired a credible institutional origin story with aggressive engineering milestones. Spun out of MIT’s Plasma Science and Fusion Center, CFS is building SPARC, a compact tokamak designed to demonstrate net fusion energy, and ultimately ARC, a commercial fusion power plant.
Fusion has long been the punchline of energy technology: always decades away. What has changed is not that fusion suddenly became easy. It is that enabling technologies improved enough to make venture-scale risk plausible. CFS’s use of high-temperature superconducting magnets is central to that case. Stronger magnets can allow smaller machines, and smaller machines can potentially shorten development cycles and reduce capital intensity.
The company’s funding history shows why investors keep showing up. CFS announced a $1.8 billion Series B in 2021, one of the largest private climate-tech rounds ever, and its continued presence near the top of funding lists underscores how fusion has moved from government lab curiosity to private-market contest.
The strategic backdrop matters. Data centers, chip fabs, electrified transport and reshored manufacturing all need reliable power. Renewables are scaling quickly, but grids still need firm, clean energy that can operate around the clock. If fusion works commercially, it does not just create another energy startup winner; it rewrites the cost curve for compute, industry and geopolitics.
That is why CFS can attract capital despite a timeline that would scare most software investors. It is a bet on abundance.
Antora’s Battery Bet Is About Factories, Not Phones
Antora Energy sits in a different part of the deep-tech map: closer to deployment, but still aimed at a hard industrial problem. Its technology is not a lithium-ion battery for vehicles or consumer electronics. Antora’s system uses surplus renewable electricity to heat carbon blocks to extremely high temperatures, storing energy as heat. That heat can then be used directly for industrial processes or converted back into electricity using thermophotovoltaic cells.
That distinction matters. Heavy industry needs heat, not just electrons. Cement, chemicals, food processing, paper, metals and other sectors often rely on fossil fuels because they require high-temperature, continuous energy. Electrifying that demand is one of the hardest pieces of decarbonization.
Antora’s reported $550 million funding package highlights how climate tech financing is becoming more sophisticated. Its $150 million Series B brought in venture and strategic investors; a separate $400 million conditional commitment from the U.S. Department of Energy’s Loan Programs Office signaled the kind of project-finance support needed to move from promising hardware to industrial scale.
This is the emerging climate-tech model: venture capital funds the company and product risk, while government loans, customer contracts and infrastructure capital help fund factories and deployments. That hybrid stack is increasingly necessary because the winners in industrial decarbonization have to build real assets, not just software dashboards.
Why These Bets Belong in the Same Story
AI, fusion and thermal batteries may look unrelated, but investors are linking them through a common macro thesis: the next decade belongs to technologies that expand capacity.
AI expands cognitive capacity. Fusion, if successful, expands energy capacity. Long-duration and thermal storage expand the usefulness of renewable power and help move clean energy into sectors that cannot run on intermittent supply alone. Each category addresses a bottleneck that could limit economic growth: intelligence, electricity and industrial heat.
This also explains why the biggest rounds are concentrating in fewer companies. Deep tech rewards credibility. Investors want teams with rare technical talent, defensible architectures, strategic partners and a believable path through regulation, manufacturing or deployment. The result is a power-law market where a handful of companies absorb enormous sums because the cost of being second-best may be very high.
There is a warning inside that concentration. Billion-dollar rounds do not guarantee billion-dollar outcomes. Frontier AI could hit scaling limits. Fusion could miss cost or timing targets. Thermal batteries could struggle with customer adoption or project economics. But the funding pattern is still meaningful because it shows where sophisticated capital believes the next chokepoints are.
Conclusion: The Next Cycle Is Physical and Foundational
The deep-tech funding cycle now forming is not a retreat from software. It is a recognition that software’s next wave depends on foundations that are harder to build: safe advanced intelligence, abundant clean power and industrial-scale energy storage.
Safe Superintelligence, Commonwealth Fusion and Antora are not interchangeable bets. But together they map the new frontier of technology investing. The most valuable startups of the next cycle may be those that do not merely optimize the digital economy, but supply the intelligence and energy it runs on.