> ## Content Index
> Fetch the complete content index at: https://www.bushletter.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Losing billions, the AI labs now want to slow down for safety
- URL: https://www.bushletter.com/losing-billions-the-ai-labs-now-want-to-slow-down-for-safety/
- Published: 2026-09-29T11:00:00.000Z
- Updated: 2026-09-29T10:59:59.000Z
- Description: Anthropic and OpenAI briefed the UN Security Council on existential AI risk while their financial disclosures show a combined $1.37 trillion in planned compute spending.
- Author: Editor
- Tags: AI, US, Anthropic

![Takeshi Mori](https://res.cloudinary.com/dz77sb7j1/image/upload/v1774262608/bushletter/authors/takeshi-mori.png)

By **Takeshi Mori** · 2026-09-29

TLDR

Anthropic and OpenAI briefed the UN Security Council on existential AI risk while their financial disclosures show a combined $1.37 trillion in planned compute spending. Anthropic lost more than $8 billion in 2025 on $4.6 billion in revenue, and OpenAI's $122 billion raise is projected to run dry by 2028.

KEY TAKEAWAYS

01Anthropic's 2025 operating loss exceeded $8 billion against $4.6 billion in revenue, per its IPO prospectus filed 28 September.

02A quarter of Anthropic's 2025 revenue came from just two customers, concentrating risk sharply.

03OpenAI closed a $122 billion funding round in March 2026 at an $852 billion valuation.

04GPT-6.1 Astra was shelved on 28 September, the same day Anthropic's prospectus published.

05OpenAI's free cash flow is projected to reach negative $278 billion by 2030 under its own financial plan.

## The verdict

Safety-first rhetoric from labs burning at this rate functions as regulatory capture. The rules they helped write raise barriers that only incumbents can clear. Read the financials.

## What the UN heard

On 23 September 2026, the United Nations Security Council convened a high-level meeting on artificial intelligence and international security. Executives from Anthropic and OpenAI briefed the Council on what they described as real and imminent threats from runaway AI development.[\[6\]](https://www.un.org/en/live-openai-and-anthropic-brief-security-council-amid-%E2%80%98real-and-imminent%E2%80%99-threat-posed-runaway-ai?trk=public%5Fpost%5Fcomment-text&ref=bushletter.com)

Five days earlier, Anthropic CEO Dario Amodei published an essay on his personal blog titled "We Must Pace the Frontier." Amodei argued that AI labs must slow the pace of capability improvements. He warned that AI-driven botnets could cause hundreds of billions of dollars in damage within six to twelve months.[\[1\]](https://darioamodei.com/post/we-must-pace-the-frontier?ref=bushletter.com) The essay landed while Anthropic's bankers were finalising an IPO prospectus that told a very different story about the company's own trajectory.

## Anthropic's numbers

The prospectus, filed 28 September 2026, showed Anthropic lost more than $8 billion in 2025 against $4.6 billion in revenue. A quarter of that revenue came from just two customers. That concentration would concern any credit analyst, let alone a safety-focused mission-driven lab.[\[7\]](https://techcrunch.com/2026/09/28/anthropics-prospectus-details-losses-growth-and-yes-a-warning-that-its-ai-could-end-humanity/?ref=bushletter.com) Run-rate revenue had surpassed $30 billion by early 2026, up from roughly $9 billion at the end of 2025\. The model is scaling fast, but costs are rising too.[\[3\]](https://www.anthropic.com/news/anthropic-amazon-compute?guides=understanding-tradeoffs&ref=bushletter.com)

| Metric                             | Figure           | Source                              |
| ---------------------------------- | ---------------- | ----------------------------------- |
| 2025 revenue                       | $4.6 billion     | Anthropic IPO prospectus, Sept 2026 |
| 2025 operating loss                | Over $8 billion  | Anthropic IPO prospectus, Sept 2026 |
| Run-rate revenue (early 2026)      | Over $30 billion | Anthropic press release, April 2026 |
| Planned compute spend              | $518 billion     | Aggregated compute agreements       |
| Revenue share from top 2 customers | 25%              | Anthropic IPO prospectus, Sept 2026 |

![Anthropic's prospectus figures: revenue, operating expenses and loss. Chart: Bushletter](https://res.cloudinary.com/dz77sb7j1/image/upload/w_1200,q_auto,f_auto/v1790669288/bushletter/charts/2026-09-29/anthropic_model.png)

Anthropic's prospectus figures: revenue, operating expenses and loss. Chart: Bushletter

The compute commitments underpin those numbers. Anthropic signed a multi-year agreement with Google and Broadcom for multiple gigawatts of TPU capacity. The infrastructure is expected to come online from 2027.[\[4\]](https://www.anthropic.com/news/google-broadcom-partnership-compute?trk=article-ssr-frontend-pulse%5Flittle-text-block&ref=bushletter.com) Anthropic CFO Krishna Rao called this part of a disciplined approach to scaling infrastructure. This is a curious word choice for a company whose losses outrun revenue by nearly two to one.[\[4\]](https://www.anthropic.com/news/google-broadcom-partnership-compute?trk=article-ssr-frontend-pulse%5Flittle-text-block&ref=bushletter.com)

A separate agreement with SpaceX gave Anthropic access to all compute capacity at SpaceX's Colossus 1 data centre, over 300 megawatts of new capacity.[\[5\]](https://www.anthropic.com/news/higher-limits-spacex?lang=us&ref=bushletter.com) That deal, signed in May 2026, followed the Google-Broadcom announcement by less than a month. Every gigawatt locked in becomes another argument against any regulator suggesting the lab should slow down. The capital is already committed.

## OpenAI's numbers

OpenAI's situation involves larger sums and steeper trajectories. The company closed its latest funding round in March 2026 with $122 billion in committed capital at a post-money valuation of $852 billion.[\[2\]](https://openai.com/index/accelerating-the-next-phase-ai/?ref=bushletter.com) Leaked financial projections show revenue scaling from $36 billion to $350 billion between 2025 and 2030\. The cash burn keeps pace. Negative free cash flow is projected to reach $278 billion by 2030\. The $122 billion raise is expected to be exhausted by 2028 under the company's own model.[\[8\]](https://www.pymnts.com/news/artificial-intelligence/2026/openais-cash-burn-could-approach-280-billion-by-2030/?ref=bushletter.com)

| Metric                                    | Figure       | Source                           |
| ----------------------------------------- | ------------ | -------------------------------- |
| Latest funding round (March 2026)         | $122 billion | OpenAI press release, March 2026 |
| Post-money valuation                      | $852 billion | OpenAI press release, March 2026 |
| Projected revenue by 2030                 | $350 billion | Leaked financial projections     |
| Projected negative free cash flow by 2030 | $278 billion | Leaked financial projections     |
| Planned compute spend                     | $856 billion | Leaked financial projections     |
| Funding exhausted (projected)             | 2028         | Leaked financial projections     |

![OpenAI's own projections to 2030: revenue, compute spend and cash burn. Chart: Bushletter](https://res.cloudinary.com/dz77sb7j1/image/upload/w_1200,q_auto,f_auto/v1790669290/bushletter/charts/2026-09-29/openai_model.png)

OpenAI's own projections to 2030: revenue, compute spend and cash burn. Chart: Bushletter

## Why the math makes slowing down impossible

Both companies have structured themselves so that deceleration is financially catastrophic. When compute agreements run to hundreds of billions and investors expect returns on an $852 billion valuation, model development does not pause because a rival lab's CEO published a blog post. The play is to ship faster and raise again before 2028.[\[8\]](https://www.pymnts.com/news/artificial-intelligence/2026/openais-cash-burn-could-approach-280-billion-by-2030/?ref=bushletter.com)

The signalling value of safety rhetoric deserves a clear read. Regulations shaped by labs at this scale, covering compute thresholds, safety-evaluation requirements, and licensing regimes, function as moats. A startup burning $10 million a year cannot satisfy the same audit requirements as a company that wrote them. Amodei's essay and the UN briefing are part of the financial plan.

On 28 September 2026, the same day Anthropic's prospectus published, OpenAI shelved the GPT-6.1 Astra model launch and apologised to Australians over a Medicare data exposure incident.[\[9\]](https://www.abc.net.au/news/2026-09-29/openai-apologises-medicare-shelves-chatgpt-astra-launch/107207156?ref=bushletter.com) Labs that brief the UN Security Council on existential risk one week and pull a major product launch the next have not resolved the tension between safety and growth. Operators building on their APIs should price that instability into their own planning. The combined compute spend across both companies sits at approximately $1.37 trillion. OpenAI's next raise, required by 2028 under its own projections, will set a new record.[\[8\]](https://www.pymnts.com/news/artificial-intelligence/2026/openais-cash-burn-could-approach-280-billion-by-2030/?ref=bushletter.com)

This article contains analysis and commentary on market conditions. It does not constitute financial, investment, or professional advice. Past performance is not indicative of future results. Always consult a qualified adviser before making financial decisions.

SOURCES & CITATIONS

1. [We Must Pace the Frontier, Dario Amodei](https://darioamodei.com/post/we-must-pace-the-frontier?ref=bushletter.com)
2. [Accelerating the Next Phase of AI, OpenAI](https://openai.com/index/accelerating-the-next-phase-ai/?ref=bushletter.com)
3. [Anthropic Amazon Compute Partnership, Anthropic](https://www.anthropic.com/news/anthropic-amazon-compute?guides=understanding-tradeoffs&ref=bushletter.com)
4. [Google and Broadcom Partnership, Anthropic](https://www.anthropic.com/news/google-broadcom-partnership-compute?trk=article-ssr-frontend-pulse%5Flittle-text-block&ref=bushletter.com)
5. [SpaceX Colossus 1 Compute Agreement, Anthropic](https://www.anthropic.com/news/higher-limits-spacex?lang=us&ref=bushletter.com)
6. [UN Security Council meeting on AI and international security](https://www.un.org/en/live-openai-and-anthropic-brief-security-council-amid-%E2%80%98real-and-imminent%E2%80%99-threat-posed-runaway-ai?trk=public%5Fpost%5Fcomment-text&ref=bushletter.com)
7. [Anthropic's prospectus details losses, growth and a warning that its AI could end humanity](https://techcrunch.com/2026/09/28/anthropics-prospectus-details-losses-growth-and-yes-a-warning-that-its-ai-could-end-humanity/?ref=bushletter.com)
8. [OpenAI's cash burn could approach $280 billion by 2030](https://www.pymnts.com/news/artificial-intelligence/2026/openais-cash-burn-could-approach-280-billion-by-2030/?ref=bushletter.com)

FREQUENTLY ASKED QUESTIONS

How much does Anthropic plan to spend on compute?

Anthropic's planned compute spend, aggregated across its agreements with Amazon, Google, Broadcom and SpaceX, totals approximately $518 billion.

When will OpenAI's $122 billion funding round run out?

Under OpenAI's own financial projections, the $122 billion raised in March 2026 is expected to be exhausted by 2028, at which point the company would need to raise again.

What did Dario Amodei argue in his 'We Must Pace the Frontier' essay?

Amodei argued that AI labs must slow the pace of capability improvements, warning that AI-driven botnets could cause hundreds of billions of dollars in damage within six to twelve months.

Is there a contradiction between safety advocacy and massive compute spending?

Critics argue yes. Both Anthropic and OpenAI publicly call for slower AI development while their financial commitments to compute infrastructure make deceleration economically impossible. Safety regulations they help design also tend to raise compliance barriers that only large incumbents can meet.

![Takeshi Mori](https://res.cloudinary.com/dz77sb7j1/image/upload/v1774262608/bushletter/authors/takeshi-mori.png)

[Takeshi Mori](https://bushletter.com/author/takeshi-mori/?ref=bushletter.com)

Takeshi Mori writes about technology and start-ups. He is curious about how products get built and who they are really for, and he would rather see a thing working than hear it described.

Important

This article contains forward-looking financial projections and estimates about AI company valuations, revenue forecasts, and cash burn. These projections are based on disclosed financial documents and leaked internal estimates and should not be relied upon as financial advice or investment recommendations. Actual results may differ materially. Readers should seek independent financial advice before making investment decisions.