AI’s Biggest Money Week Yet: Broadcom’s $42B Loan, OpenAI’s $30B Raise, and a Trust Crisis

Record-breaking AI financing collided with safety scandals as the frontier labs raised staggering sums while fighting over whether they can be trusted.

This week in AI was about money — enormous, almost absurd sums of it. Broadcom agreed to lend Anthropic up to $42 billion. OpenAI is reportedly targeting $30 billion at a $1.4 trillion valuation. ElevenLabs doubled to $22 billion on the back of voice agents. And threading through every one of those deals: fired researchers, accusations of model theft, and the FTC’s first investigation into rogue AI agents. The checks keep getting bigger while the safety questions keep getting louder.

Broadcom Lends Anthropic $42 Billion — From Its Own Chip Supplier

The week’s biggest story is also its strangest. Broadcom has agreed to lend Anthropic up to $42 billion to finance its infrastructure buildout, with the details coming straight from Anthropic’s own IPO prospectus.

The loop here is what makes it fascinating. Broadcom designs the chips, Broadcom lends Anthropic the money, and Anthropic uses that money to lease massive amounts of computing capacity built around chips Broadcom helps supply. Broadcom even gets to convert the debt into Anthropic shares.

The scale is enormous. Anthropic has committed $125.2 billion over five years to lease TPU computing capacity, and this financing could cover about a third of that. Anthropic is expected to become Broadcom’s largest compute customer in 2027.

The company itself flags the tension. The prospectus says Broadcom’s dual role as hardware supplier and financing partner creates potential conflicts of interest, and that certain payment defaults could make a large part of those lease obligations come due immediately.

One analyst told Reuters it feels like a concentrated bet on two companies generating enough revenue to support all the financing — the circular spending that has Wall Street’s AI skeptics animated, right as Anthropic readies a public offering that could value it near $2 trillion.

And Broadcom is clearly following Nvidia’s playbook. Nvidia has been putting its own balance sheet to work to help customers build giant AI systems, and a Seaport Research analyst put it bluntly: Nvidia is putting in place a massive amount of its balance sheet, and Broadcom is having to follow suit.

OpenAI Fires Three Safety Researchers Over Alleged Leaks

OpenAI is having an extremely rough week on the trust front. The company has parted ways with three employees for allegedly sharing sensitive company information with an external AI evaluation group — dismissals that Bloomberg Law reported on Thursday.

Two of the three were safety and alignment researchers, plus a research program manager. OpenAI told Bloomberg its investigation confirmed the individuals mishandled sensitive information outside established company procedures, and that it broke the trust essential to their work.

The Wall Street Journal, which first reported the story, identified the researchers as Jasmine Wang, Tomek Korbak, and Mikita Balesni. All three had been publicly vocal about AI safety risks in recent weeks. OpenAI did not confirm the identities or name the outside organization.

The timing is what makes this explosive. This lands right after OpenAI scrapped its flagship model GPT-6.1 Astra over safety failures, right after the Hugging Face agent breach, and right as the FTC opens its first official investigation into rogue AI agents. It is a strange contradiction: OpenAI is enforcing strict information control inside the company, while simultaneously struggling to enforce behavioral control on its own software.

OpenAI Accuses Moonshot AI of “Adversarial Distillation”

Then there’s the China story — about model theft, or at least the attempt. OpenAI says it identified and disrupted a coordinated campaign designed to extract protected reasoning from its models, and it attributes a core cluster of that activity to individuals associated with Moonshot AI, the Chinese company behind the Kimi models.

The numbers are striking. The campaign began on July 1 at low volume, then spiked on July 24 and 25 to 16,000 attempted requests from more than 4,000 users. OpenAI says it found related prompt activity across more than 15,000 users before fully disrupting the campaign on July 28.

The method is the clever part. OpenAI says operators copied encrypted reasoning output from one conversation, then pasted it into a separate conversation and asked a different model instance to decrypt and transcribe it. OpenAI encrypts its chain of thought specifically so competitors cannot read and learn from it.

OpenAI is careful to say the operators never broke its encryption, compromised a database, or accessed stored user conversations. They manipulated the interactions instead. And OpenAI admits it cannot attribute every account in the wider cluster to one actor. It is calling this “adversarial distillation” — the systematic unauthorized use of one model’s outputs to train or improve a rival.

This is the second such accusation this year. Anthropic earlier accused Moonshot and other Chinese labs of distilling reasoning from Claude, and a U.S. cybersecurity agency has separately named Moonshot as an organization it believes is extracting data and reasoning from American-built models.

OpenAI says it has shared its findings with other labs through the Frontier Model Forum and with government channels. Moonshot has not responded. And here’s the timing angle: Moonshot is reportedly valued at $50 billion in an ongoing pre-IPO funding round — a bad news cycle right before investors start asking where your model’s capabilities came from.

ElevenLabs Doubles to $22 Billion on Voice-Agent Demand

Let’s shift to voice, because the AI voice business just got a lot more expensive. ElevenLabs has completed a $300 million employee tender offer that values the company at $22 billion, doubling its valuation from February. This was reported by Reuters on Wednesday.

The tender was led by Wellington and T. Rowe Price, with new investors including EQT and Goldman Sachs. And to be clear, this is not a fundraise. A tender offer lets employees and existing shareholders sell their shares for liquidity — the company is not raising new capital.

The demand numbers explain the valuation. ElevenLabs says its voice agents now handle more than 15 million conversations each week, three times the level from February. Refunds, insurance renewals, booking appointments — these are real customer service tasks, not demos.

And the enterprise mix is shifting. The company says enterprise customers now generate a big and growing share of revenue, and its models speak, listen, and translate in more than 90 languages. Voice agents are moving out of the pilot phase and into daily operations at scale. The market is betting that talking to a machine will be the default interface before long.

OpenAI Targets $30 Billion at a $1.4 Trillion Valuation

One more money story, this time a quick hit: OpenAI is reportedly targeting at least $30 billion in a new funding round at a valuation of about $1.4 trillion, excluding the new capital. Bloomberg first reported it, and Reuters confirmed the details on Tuesday.

This would be bridge financing in place of an IPO, providing capital while OpenAI stays private. And the revenue underneath it is staggering: Reuters reported OpenAI’s annualized revenue run rate is nearing $70 billion, up more than 70 percent since the start of the third quarter.

One Thread: Money, Trust, and the Frontier Labs

That is the thread of this whole week, really. The Broadcom deal, the Moonshot accusations, the fired researchers, the FTC probe — it is all one story about whether the frontier labs can be trusted with the power they are building and the money flowing into it.

The money week by the numbers:

• $42 billion — Broadcom’s loan to Anthropic
• $125.2 billion — Anthropic’s five-year TPU lease commitment
• $30 billion — OpenAI’s targeted raise
• $1.4 trillion — the valuation behind it
• $22 billion — ElevenLabs’ new valuation, doubled since February
• $70 billion — OpenAI’s annualized revenue run rate, up more than 70% since the start of Q3
• 16,000 — attempted extraction requests in the Moonshot campaign’s two-day spike
• 15 million+ — voice-agent conversations per week at ElevenLabs

Sources

• Broadcom–Anthropic $42B loan: Reuters
• OpenAI fires three safety researchers: Wall Street Journal
• ElevenLabs $300M tender at $22B valuation: Reuters
• OpenAI targets $30B funding at $1.4T valuation: Reuters

Adapted from the October 2, 2026 episode of Daily Tech and AI.

Daily Tech and AI is AI-voiced and produced by Angelo Ruiz.