OpenAI Wants $1.2 Trillion — While Burning $278 Billion

OpenAI is out shopping for money again, and the numbers have officially left the atmosphere. According to a Financial Times report published Friday, the company behind ChatGPT is in early talks for a new funding round at a valuation above $1.2 trillion — even as its own investor presentation projects a $278 billion cash shortfall between now and the end of 2030. In Indian terms, that gap is over ₹23 lakh crore. At some point these numbers stop feeling like money and start feeling like weather.

The $278 billion question

Here's the shape of it, per the presentation reviewed by the FT. Between 2026 and 2030, OpenAI expects to bring in roughly $840 billion in total revenue. Its spending on computing power and infrastructure — its single largest expense — is expected to reach $856 billion. That leaves a projected negative free cash flow of $278 billion, though it's actually an improvement on the $305 billion shortfall the company projected in May.

Why does it cost so much? Because the model everyone uses for a few hundred rupees a month runs on data centers that cost tens of thousands of crores to build and operate. Every conversation you have with ChatGPT passes through Nvidia GPUs that eat electricity like a small town. OpenAI is betting that if it builds enough of this capacity first, it owns the layer everyone else has to rent.

Revenue is growing fast — just not fast enough

The revenue line is genuinely wild: from $36 billion in 2026 to a projected $350 billion in 2030, nearly a tenfold jump. The FT report notes annualized revenue rose about 20% in July alone following new model releases. By any normal company's standards, this is one of the fastest-growing businesses in history.

So what's the problem? Prices keep falling. The report says OpenAI has been cutting prices to compete with rival Anthropic and with cheaper open-weight models from China — and when your revenue grows tenfold while your unit prices fall, you need volume at a scale nobody has ever seen. Honestly, what surprised me most is how openly the deck admits the March money could run out by 2028. This isn't a company hiding its burn rate; it's a company selling its burn rate as the plan.

Why investors might pay $1.2 trillion

In March, OpenAI raised $122 billion at an $852 billion post-money valuation — Amazon, Nvidia and SoftBank anchored the round, Microsoft continued participating, and SoftBank co-led alongside Andreessen Horowitz, D.E. Shaw Ventures, MGX, TPG and T. Rowe Price. Now it's asking for more than $1.2 trillion, a 41% markup in under six months, while the same documents suggest the previous pile lasts only until 2028.

Why would anyone sign up for that? Because they're already locked in. Nvidia, Oracle and SoftBank's data center business rely heavily on OpenAI contracts, as the report notes — if OpenAI grows, their order books grow with it. We saw the same pattern when the $105 billion Activate AI Fund launched to back AI infrastructure earlier this month, which we covered here. The AI economy is increasingly a web of companies financing each other's bets.

The IPO that got iced

OpenAI had confidentially filed IPO paperwork in June, targeting a fall listing. That's off for now — Altman said on September 12 that OpenAI won't go public in 2026, arguing that staying private gives the company room to focus on AI safety (though, as the FT notes, some investors also worry how markets would receive a heavily loss-making listing). Meanwhile, rival Anthropic is expected to go public this fall, and it's reportedly weighing a new model launch ahead of the listing — we broke down Anthropic's IPO plans here.

So the money story and the safety story are now the same story. The most valuable private company on earth is telling investors: give us a trillion dollars, and also, we need to slow down for safety. Whether both can be true at once is the question of the decade.

What this means for India's AI economy

For Indian founders, freelancers and students, there's a very practical lesson hiding in these numbers. That cheap API access you enjoy — the subsidized tokens, the price cuts — is being paid for by investors, not by sustainable economics. Build your product, your service business, your AI workflow on the assumption that prices eventually have to cover costs. The smartest Indian AI businesses we see are the ones treating cost optimization as a core skill, not an afterthought.

That's exactly why our GPT-6 Astra — Zero to Hero Guide spends a full chapter on cost optimization — prompt caching, model selection, and workflows that keep your monthly bill survivable. The companies that master this math are the ones still standing when the subsidy era ends.

Key Takeaways

• OpenAI projects a $278 billion cash shortfall from 2026 to 2030 — $856 billion in compute spend against $840 billion in revenue
• Revenue is projected to grow from $36 billion in 2026 to $350 billion in 2030
• A new round is being discussed at a valuation above $1.2 trillion, up from $852 billion in March
• The $122 billion raised in March could run out by 2028; the IPO is deferred while Anthropic heads for a fall listing
• For Indian builders: cheap AI access is subsidized — treat cost optimization as a survival skill

Frequently Asked Questions

What This Means for You

If you're a student or professional betting your career on AI skills, this news is quietly encouraging — the biggest company in the field is spending a quarter of a trillion dollars to make these tools cheaper and more capable for you. If you're building a business on top of them, the lesson is the opposite: don't build on subsidized prices forever, and learn the cost side of the equation cold.

Either way, the next two years reward people who actually understand these tools deeply. The GPT-6 Astra — Zero to Hero Guide is our pick for getting there fast — prompting, agents, computer use, cost optimization, business ideas and safety, all in one place.