OpenAI has cut its funding target by a third. The investors still writing cheques want revenue, and that will show up in your contract.

OpenAI has cut the target for its latest funding round from $30 billion to $20 billion, the Financial Times reported on 30 October, and lowered the valuation it is asking for from $1.4 trillion to $1.25 trillion. A group of investors from the United Arab Emirates leads the round, and the first close is expected to come in below the new target, TechCrunch reports. According to the Wall Street Journal, OpenAI had secured about $11 billion so far, and some large investors gave the round what the paper called a tepid response.

OpenAI did not comment. $20 billion is still an enormous private raise, and the company is not short of money. But the cut tells you what investors now want from the AI labs, and buyers will feel it.

Why OpenAI lowered its target

Timing did most of the damage. Investors had until 30 October to commit, and that Friday the Nasdaq fell 2.9%, leaving it 7.8% below its high of 6 October, according to Reuters figures summarised by Ground News. The day before, Alphabet fell about 12%, its worst day on record, after its cash flow went further negative on data centre spending, Fortune reports.

The same evening S&P Global cut its outlook on Meta to negative. It expects Meta to spend about $140 billion on capital projects this year and more than $225 billion in 2027, and to burn cash for three years running, and said it was unclear when the AI products would pay for it. Meta disagreed.

Anthropic is the other reason. Its listing on Nasdaq is expected in mid-November, and Reuters reported on 20 October that OpenAI wanted its round closed before Anthropic’s prospectus gave investors a fuller comparison. The two companies court the same funds. We covered what that prospectus showed in Anthropic’s leaked IPO filing and in its later admission of a month-long training pause.

What the numbers say about OpenAI’s position

On paper, OpenAI’s value is still rising. A valuation of $1.25 trillion is almost 50% above the $852 billion it was valued at in March, when it raised $122 billion, as Wealth Professional notes, citing the FT. A person close to the company told the FT a stock market listing is likely in the first half of 2027.

The pitch to investors rests on revenue growth. OpenAI told them it had reached $50 billion in annualised revenue and expected $70 billion by the end of the year, according to Reuters. It also said it had cut prices on its cheaper models to drive usage and could cut further.

A company that lowers prices to show investors faster growth is buying market share, and the bill for that comes due once growth has to turn into profit. The two labs are already arguing over how to count revenue: Reuters reports that OpenAI accuses Anthropic of inflating its figures by counting the full value of contracts sold through cloud providers, which Anthropic denies.

The skeptics and the bulls

Not everyone reads this as trouble. The market strategist Ed Yardeni told clients the financial, physical and political limits now biting are “healthy for the durability of the AI boom”, Fortune reports, because tighter money curbs overbuilding. He expects the big cloud companies to keep spending and sees further sell-offs as buying chances.

The other view is that this is the first time AI money has been hard to raise. Rockefeller Global Family Office’s investment chief, who holds stakes in both labs, told the FT that the market needs to see a successful IPO. Anthropic’s listing is now carrying the mood of the whole sector. If that goes badly, OpenAI’s 2027 plans get harder too.

I side with Yardeni on the outcome and with the worriers on the mechanism. Neither lab is about to run out of money. But both now answer to investors who want proof that revenue can cover a trillion-dollar valuation and hundreds of billions in compute commitments, and the place that proof comes from is enterprise customers.

What European companies buying AI should do now

Expect your account manager to push for bigger, longer commitments. Labs under pressure to show growing revenue prefer multi-year contracts with spending floors, because those make the numbers predictable. If you sign one, negotiate exit terms, price protection for the models you depend on, and the right to move spend between models as new ones arrive. A steep fall in token prices is little use if you are locked into last year’s rate.

Take advantage of the price competition while it lasts, but do not plan on it. OpenAI has told investors it can cut prices further, so today’s rates may fall again. Build your business case on current prices and treat any further drop as a bonus.

Keep a second supplier ready. Most companies that use AI at scale lean on one or two American labs. Make sure your core workflows can switch to another model within a few weeks, and test that switch once a year. European options such as Mistral, and the capacity Brussels and Copenhagen are now paying to build in Europe, are worth keeping on the list.

Ask where your data is processed. The FT reports that the new round gives OpenAI first refusal on new computing power built by its UAE partners. That may never touch your workloads, but under GDPR you need to know which countries your data passes through. Check that your contract fixes the processing region, and that any change requires your consent.

Then read Anthropic’s prospectus when it becomes public. It will be the most detailed account yet of what a frontier lab earns, spends and owes, and it will tell you more about the stability of your supplier than any sales deck.