Fusion research pushes toward the grid as Singapore, UK and California move in

Singapore's interest was reported by CNA and the Business Times on 1 October, the same day a fusion conference in the city-state put spin-offs and supply chains on the agenda. Manpower Minister Tan See Leng said fusion could create economic opportunity for Singapore, according to The Straits Times on 30 September.

The trade ministry was renamed in a move IndexBox linked to the sector's growing profile. The city-state is not building a plant. Its government is studying nuclear power, including small modular reactors, fusion investments and financing, according to Nuclear News Network's topic page. Most of the movement in fusion right now is money, regulation and engineering, not electricity.

Money and rules arrive before the electrons

The UK committed £2.5 billion ($3.3 billion) to commercialise fusion energy, a figure reported by ESG News on 30 September. The UK's Health and Safety Executive also published new guidance for fusion facilities, according to HazardEx. California signed fusion legislation and announced a $30 million quantum investment, The Quantum Insider reported on 1 October, with Governor Newsom's office framing the package as state capital for quantum infrastructure.

Private capital is moving in parallel. Commonwealth Fusion Systems raised another $1 billion in equity, lifting total capital to $4 billion as assembly of its SPARC machine nears completion, per Nuclear News Network. Japan's METI conditionally selected Helical Fusion and three others for a JPY 60 billion fusion programme, with a stellarator demonstration targeted at the 2030s. Tennessee issued the first fusion-specific state licence in the US, clearing Type One Energy to break ground on a stellarator campus at Bull Run.

Set against those commitments, the physics record in the dossier is older. The Joint European Torus near Abingdon in Oxfordshire generated around 59 megajoules, or 11 megawatts, in a five-second burst, according to London Daily. The machine's plasma reached about 150 million degrees Centigrade, roughly 10 times hotter than the sun. JET still consumed more energy to create the reaction than the reaction released. Professor Ian Chapman, chief executive of the UK Atomic Energy Agency, called the results a "landmark", and Professor Ian Fells of the University of Newcastle said the same. The dossier does not date the JET run, and it should be treated as background rather than news.

A density claim from China, and a theory that predicted it

The newest experimental claim in the dossier comes from China. Researchers at the Experimental Advanced Superconducting Tokamak, EAST, say they reached a "density-free regime", pushing plasma density past a limit that has constrained tokamaks for decades without triggering the instabilities that normally shut a shot down.

The work was co-led by Professor Ping Zhu of Huazhong University of Science and Technology and Associate Professor Ning Yan of the Hefei Institutes of Physical Science at the Chinese Academy of Sciences, and published in Science Advances, according to Custom Map Poster. That outlet frames the result as support for plasma-wall self-organization theory, pioneered by D.F. Escande at the French National Center for Scientific Research and Aix-Marseille University. The reported mechanism is control of physical sputtering, the process where plasma bombards the reactor wall and releases particles that change plasma behaviour. Fusion power scales with the square of plasma density, so a higher stable density is a direct route to more reactions. Treat the claim with care: it is a single outlet's account of a journal paper, and no independent replication appears in the dossier.

Elsewhere in China, Digitimes reported on 1 October that the BEST fusion reactor has entered an engineering endgame, and Sada Elbalad reported on 28 September that the country completed a major superconducting magnet for a fusion project. The South China Morning Post reported on 28 September that a Chinese commercial reactor achieved fusion with hydrogen-boron fuel, and Nature World News said ENN broke ground on a hydrogen-boron device, alongside a House bill proposing $10 billion for fusion energy.

The AI research wave reaches the lab bench

Fusion is not the only field where automated systems are being pointed at hard science. Microsoft Research introduced Quine on 29 September, described as a multimodal world model of biology paired with an interactive harness connecting models, tools, literature and researchers. Working with the Broad Institute of Harvard and MIT, Microsoft says the system prioritised compounds predicted to drive therapeutic tumour-state shifts and validated several top-ranked candidates across wet-lab assays. The company calls it experimental research technology, not for clinical use, and says outputs may be incomplete or inaccurate.

The same week, CoreWeave launched ARIA, a coding agent built into Weights and Biases that reads experiments, forms hypotheses, launches runs and drafts reports. Efficient Computer announced a $97 million Series B led by TQ Ventures for energy-efficient processors, with CEO Brandon Lucia arguing that Amdahl's Law caps the gains from narrowly specialised accelerators. An NBER working paper by Matthew Schwartz, Isaiah Andrews and Jesse M. Shapiro describes an LLM workflow that ran across 4,452 published replication packages in five economics journals and flagged discrepancies in 3,460 articles or their appendices.

Disclosure matters here: the NBER paper states that Schwartz worked as a contractor for Anthropic and that the results and views are not endorsed by Anthropic. The authors also note LLMs were used in the analysis and writing of the article.

Two mathematicians pushed back on the idea that AI makes human researchers redundant. Stephen Wolfram wrote on 28 September that AI is "unquestionably useful" for mining the mathematical literature but that pure mathematics is not reducible to producing proofs. Dan Romik argued on 29 September that mathematics research is a feedback loop, and that models trained on the existing corpus can generate theorems "distance one away" from human knowledge but will stall without the ability to reflect on, simplify and connect what they produce. Romik also quoted Scott Aaronson's reaction to OpenAI's claimed solution to the Navier-Stokes problem, a string of A's.

Energy demand is the other half of the story

Data centres and electrification set the demand side. Newfoundland and Labrador's energy minister, Lloyd Parrott, said the province's door is "open for business" after being approached by companies about AI data centres, CBC News reported on 29 September. A departmental statement said data centres may be considered where proposals align with policy objectives. Tech journalist Paris Marx, author of Hyperscale, urged caution in the same piece.

Hannah Ritchie wrote on 29 September, citing work by Oxford professor Nick Eyre, that global final energy demand could fall about 40% after the transition, from 416 to 247 exajoules, even as electricity demand rises from 110 to 189 exajoules. That arithmetic is one reason fusion's backers keep arguing the prize is worth pursuing, as Professor Sue Ions of the Royal Academy of Engineering put it in the London Daily account.

The counterweight is jobs and policy. The US lost 36,949 clean energy jobs in 2025, the first annual decline since the pandemic, according to E2's analysis of Department of Energy data reported by Electrek on 28 September. Clean energy employment fell to 3.52 million, erasing nearly 40% of 2024's gains, with E2's tracker recording 142 cancelled or downsized projects.

For fusion specifically, the newest developments are financial and regulatory, not physical. A record 59 megajoules from JET, a contested density result from EAST, and licences, legislation and billion-dollar rounds elsewhere. The dossier contains no demonstration of net energy gain, and no dated commercial timeline.