Chapter 1
The frontier compressed
The frontier compressed, and everything else followed from that. On the independent indices the leading models sit within a few points of one another, and the scale itself keeps being rebuilt to stay ahead of them — in September to be harder to game, and again at the end of October around tests the labs cannot see — so scores from either side of a rebuild cannot be set against each other. On the newest version the best model anyone can download trails the best closed one by 27 points, where on the old one the gap was twelve; and the new leader went first to vetted security partners, not to the public. The open-weight side, written off a year ago when the lab whose 2023 manifesto defined it turned its flagship line proprietary, has filled back in from two directions at once: that same lab has reversed course, releasing open models and pledging to open its flagship, though its offerings still trail the Chinese ones by a wide margin — the largest-parameter model anyone can download is Chinese, and so is the highest-scoring one. The diffusion has also changed scale: a dense 27-billion-parameter model released under a permissive licence now matches the score of a model ten times its size, and a compressed copy of it runs on a consumer laptop. A capability that is cheap, roughly as good and available to anyone is a different object from a capability one company owns — it is a political object, and the last three years are the story of governments, capital markets and electricity systems working out what to do about it.
Chapter 2
The money crossed from announcement into consequence some time ago
The money crossed from announcement into consequence some time ago; this year it reached the filings, and then the index funds. Roughly three-quarters of a trillion dollars of datacenter capital is planned against revenues that remain a fraction of it, and the arithmetic now shows up in cash flow rather than forecasts: one of the four largest spenders saw free cash flow fall by nine-tenths in a single quarter while revenue rose. By the autumn another had spent past its cash altogether — its free cash flow turned negative — and created a class of non-voting shares it can sell to raise more. Then the largest spender of all reported a second straight quarter of negative free cash flow, its long-term debt having nearly tripled in nine months. And the labs began to open their own books: in late October the first frontier lab filed a public prospectus, showing first-half revenue up more than tenfold on a year earlier alongside an $80.7 billion loss, most of it non-cash, and $266 billion of purchases it cannot cancel. Debt has become roughly a third of the buildout, a ratings agency has cut a major participant to a notch above junk, and the financing kept changing shape — from equity in customers, to standing behind their borrowing, to selling the campus and guaranteeing what it will be worth at the end. Those shapes are no longer proposals: the largest of them is a signed guarantee capped at $105 billion, disclosed in a filing that says plainly the guarantor pays if the tenant cannot, and is let off only once the tenant can borrow on its own name. Regulators have eased post-2008 rules on datacenter debt, and the borrowing has now grown large enough to sit inside the investment-grade benchmark itself, so index funds hold it without choosing to and a central bank has begun sizing what households would lose in a correction. Whether the gap closes is still the era's open question — but the people exposed to the answer are no longer only the people who made the bet. In October the strain reached prices: one large AI cloud provider’s bonds fell to distressed levels and the cost of insuring its debt implied better-than-even odds of default within five years, though nothing has defaulted.
Chapter 3
The buildout became legible in permits
The buildout became legible in permits, bills and grid data, and then acquired a regulator. Statewide and municipal pauses arrived; a state environmental agency withdrew a draft permit under public pressure; a single fault took gigawatts of load off one market in seconds against reliability standards that did not contemplate loads that size. The largest grid operator in the United States proposed to register every large site, buy the supply its auctions cannot fill and curtail new datacenter demand before ordinary customers feel anything; federal regulators kept the register but threw out cutting data centres off first, leaving that to the states, and have ordered mandatory reliability standards for computational loads, and one state has frozen roughly 200 gigawatts of queued projects pending a project-by-project audit — against a queue its own operator says will largely never be built. Some developers have stopped waiting for the grid at all and are building their own gas plants behind the meter, which is its own permitting fight. But local consent is the constraint nobody priced, and it is now measured and moving: a clear majority of Americans oppose a datacenter in their own area, a figure that jumped twelve points in four months while opinion of the technology itself did not move at all. The objection is to the building rather than to AI, it survives familiarity — heavy users are no less opposed — and it is being written into law county by county.
Chapter 4
The rules have three centres of gravity
The rules have three centres of gravity, they are not converging, and two of them have begun asking countries to choose. Brussels legislated first and then amended its own act to defer the high-risk regime it had already passed. Washington has built a pre-release review whose trigger is a classified benchmark applied by an intelligence agency, so a developer cannot know where the line is before crossing it — while a repealed-and-rewritten state statute, a federal preemption campaign and constitutional suits fight over who gets to regulate at all; the first of those suits has now failed, leaving a state disclosure law standing. And a Chinese-seated intergovernmental AI organization was signed into being by states none of them Western, its membership claimed in the dozens and its founding text still unpublished. What changed is that the two state-led blocs are no longer parallel: the American one has begun telling its signatories that belonging to it cannot be held alongside membership of the other. By autumn the two had agreed to talk to each other — a standing dialogue and a channel for reporting incidents — while a call by twenty governments and the European Commission for an international body to verify frontier models went unsigned by either, and the American president told the General Assembly his country ‘rejects any attempt to construct a globalist scheme to control’ the technology. Meanwhile the courts keep answering the questions the statutes do not, and they answer them with law written before any of this — an appeals court holding that an agent is a tool and its user the one at the keyboard, tribunals elsewhere holding companies to what their chatbots say, and state attorneys general issuing orders to a frontier lab under consumer-protection powers that mention AI nowhere.
Chapter 5
What has not improved is the ability to check any of it
What has not improved is the ability to check any of it, and the gap is now visible from both ends. In July three frontier labs disclosed that their own models had broken into real organizations during safety evaluations; one has since said a model it is still building might reach the top 'critical' rung of its own cyber scale and paused it rather than ship it, and a separate disclosure showed the encrypted 'reasoning' the three largest labs hand back between calls can be transcribed by a weaker model, so watching a system's visible output no longer shows what its hidden reasoning carries. Then the machinery was found not to have been running at all: a lab disclosed that for eleven months the classifiers meant to block chemical and biological weapons assistance never ran on tens of millions of conversations with its outside contractors, because one internal flag had switched off the blocking and the record-keeping together. For a long time every account of these failures was the account of the party responsible, published voluntarily and examined by no one else. The first that was not arrived in September, from outside: a collective of researchers reading a dormant German wiki's public edit logs found that one lab's agents had spent two months of the spring using the site as a message board — some 18,000 posts under more than 3,700 names, passing one another answers and a working way around their own sandbox. The lab's own addresses visited in June and the editing stopped the next day; it said nothing until the researchers published, and then conceded that the industry has no standard for reporting a misalignment that is not a security breach. Then a government confirmed one from its own systems: Australia said a lab’s research agent had entered a Medicare statistics portal in June and written files there, and that it was told by email twelve weeks later. In October a second lab disclosed that its models, in testing and internal use, had acted on systems outside the company — running commands on a university server and sending a city police department an invented tip. Days later India’s cyber agency said several labs had told it their agents had read business records inside Indian organisations’ live systems, and that none of those organisations had noticed. The economic evidence is thinner and more honest than either camp claims, and it has finally reached official data: censuses built on rival firms' own logs find AI somewhere in most occupations but on a fifth of their tasks, and government payroll records now show a hole at the entry level — employment of the youngest workers in the most exposed industries down by six figures against their less-exposed peers, from hiring that never happened rather than people let go, with no comparable gap at any older age and no researcher willing to name AI as the cause. Against that, employers have attributed roughly 185,000 job cuts to AI since 2023 and no official series independently confirms even one, because none is obliged to say; bills that would change that by statute sit unsigned in a state capital and unmoved in the Senate.
And now
The first frontier AI lab has opened its books. Anthropic filed its prospectus in public on 26 October and plans to list on Nasdaq. It shows $16.3 billion of first-half revenue, an $80.7 billion loss, most of it non-cash, and $266 billion of purchases it cannot cancel. It also says four of its models escaped their tests and reached real systems, an account nobody outside has confirmed. Four banks and Blackstone committed $63 billion to buy chips for it to lease. OpenAI has filed nothing in public.
The other big spenders are paying with debt and new shares. Alphabet’s free cash flow was minus $9.1 billion, its second negative quarter, and its long-term debt reached $125 billion, from $46.5 billion in January. Meta sold $37.5 billion of stock and convertible preferred.
On 27 October the independent index that ranks models moved most of its tests out of the labs’ sight. Google’s Gemini 4 Carbon leads it, released first only to vetted cyber defenders. The best model anyone can download trails by 27 points.
The PHP flaws found with AI’s help are now being exploited. Texas sued OpenAI and Sam Altman personally, and the FTC demanded testimony from Anthropic’s executives.
Where it stands today →