(Source: Shutterstock)
This week could have been an episode of HBO’s “Silicon Valley” as billions of dollars flooded into AI firms almost overnight.
The tone was set on Friday, when Nvidia became the first public company to reach a $4 trillion market cap, up from $1 trillion in 2023. By packaging its GPUs with proprietary software and cloud-scale systems for all-inclusive AI development, the company has launched an unprecedented payoff for its shareholders.
But even companies with zero product (Pied Piper, anyone?) are soaking up waves of new funding. Just take a look at Thinking Machines Lab, a company founded and led by former OpenAI CTO Mira Murati, one of our AIwire People to Watch. The startup has just closed a staggering $2 billion seed round led by Andreessen Horowitz, boosting its valuation to $12 billion before a single product has seen daylight. Even in a market awash with AI cash, a seed round of that scale is almost unheard of, signaling extreme confidence in Murati’s track record and in the promise of another frontier model contender.
The investor lineup reads like a who’s who of AI hardware and software power players: Nvidia, Accel, ServiceNow, Cisco, AMD, and Jane Street joined the round. Reports only a month ago predicted the valuation at $10 billion, so the final figure shows how quickly expectations and deal terms are inflating. Thinking Machines is less than a year old and has yet to detail its research, but Murati told followers on X that an unveiling is “a couple months” away and will feature a “significant open source offering” aimed at researchers and startups building custom models.
“Soon, we’ll also share our best science to help the research community better understand frontier AI systems,” Murati wrote.
Thinking Machines Labs CEO Mira Murati
These hints of open sourcing have fueled speculation that Thinking Machines may release its own model to challenge ChatGPT’s dominance. Murati has already recruited several OpenAI alumni, including John Schulman, Barret Zoph, and Luke Metz, and the company is hiring engineers with a history of taking AI products from idea to production. Meta even explored acquiring the startup earlier this year, which is a reminder that in today’s AI frenzy, promising labs can turn into coveted assets in a blink.
With $2 billion in the bank and a cloud deal with Google to power training runs, Murati now faces the same hurdles as her much larger rivals: scientific breakthroughs and commercial traction. Whether this record seed round buys enough time and compute to clear those hurdles is the question every AI investor will be watching.
If Thinking Machines’ $12 billion estimated value raised eyebrows, the whispers around Anthropic are dropping jaws. According to The Information, some investors are discussing a fresh round that would value the three-year-old lab at more than $100 billion, a 60-plus-percent jump from the $61.5 billion Series E Anthropic closed just four months ago. Numbers that large look less speculative when set beside Anthropic’s recent growth curve. Analysts at Sacra estimate the company is now running at $4 billion in annualized revenue, nearly quadruple its pace at the start of 2025, even as cash burn remains high.
(Source: Andrey Suslov/Shutterstock)
Amazon’s reported plan to deepen its stake beyond the $8 billion it has already committed adds another signal that major cloud providers see Anthropic’s Claude models as strategic infrastructure, not just an experimental chatbot.
Whether that $100 billion talk becomes a signed deal or just another rumor, it captures a market mood in which leading AI labs are being priced more like oil companies than software startups. The sky-high $100B valuation lands amid a high-stakes recruiting arms race. Meta’s new Superintelligence Labs and OpenAI have been trading researchers with signing bonuses in the hundreds of millions, and Anthropic has been hustling to retain and attract engineers of its own.
Speaking of the recruiting arms race, another eye-popping story is the saga of Windsurf (formerly Codeium), the startup headquartered in the exact Mountain View office complex that HBO used as Pied Piper’s set.
Late last week, Google struck what amounts to a $2.4 billion “reverse-acquihire”: it paid a licensing fee, without taking equity, to gain Windsurf’s code-generation technology and hire away CEO Varun Mohan, co-founder Douglas Chen, and a slice of the R&D team for DeepMind’s agentic-coding push. The deal came together just hours after an earlier $3 billion offer from OpenAI reportedly expired, underscoring how frenzied the bidding for elite AI engineers has become.
(Source: doomu/Shutterstock)
With much of its senior talent gone, Windsurf’s remaining 200-plus employees and its intellectual property changed hands again within 72 hours. Cognition, the startup behind the autonomous software engineering agent Devin, announced on Monday that it will acquire the rest of Windsurf, guaranteeing all staff a financial stake. In effect, Google bought the brains, Cognition bought the codebase, and investors watched a young company get broken up into parts worth more than its whole. The exact financial details were not disclosed, but the two rapid deals show how easily a company can change hands when leading AI researchers are the main asset.
Read the week’s news as a whole and one theme stands out: money is moving faster than software, just like in the dotcom boom. Markets now place a $4 trillion price on the hardware stack, billions on labs that have yet to disclose a product, and even more billions on teams hired for their next idea. The price tags have grown so outsized that the line between boardroom reality and TV satire is hard to see. “Silicon Valley” once felt like parody but now reads like yesterday’s news … and the season finale is still a long way off.