The Wall Street Journal reported on 10 October 2026 that Google co-founder Sergey Brin “sometimes upends the formal compute allocation process, reshuffling power around to projects that he thinks are important,” citing people familiar with the matter.
The report, by Kate Clark, Erin Woo, Keach Hagey and Meghan Bobrowsky, says researchers inside Google are “fiercely divided over which projects deserve the company’s precious computing resources.” It sets that account inside a wider piece on how scarce computing power has become across the AI industry.
The WSJ says the divisions sit inside Google’s Mountain View offices. Alongside the account of Brin, it reports that Demis Hassabis “has expressed frustration about a lack of compute allocated toward long-term research,” also citing people familiar with the matter.
The paper describes Hassabis as the co-founder of Google’s AI lab DeepMind, who until recently ran it and took a chairman role in August. It does not say which projects gained or lost compute when Brin reshuffled it.
Six compute moves the Wall Street Journal reports on 10 October 2026; the three it places inside Google are highlighted. Grouping and count are ours.
Business Pill · RELAYED REPORT
A one-minute explainer of the relayed report: how to read a story that rests on a publication’s anonymous sources. It teaches the general idea only and says nothing about any company in this story.
The key insight: As we read it, the WSJ describes compute at Google as something that is allocated through a formal process and, at times, reallocated from the top: it reports Brin reshuffling capacity, Hassabis frustrated about compute for long-term research, and Shazeer telling a team its allocation had been cut. The report does not give the size of any of those allocations.
The Shazeer Note
The report opens with Noam Shazeer. The WSJ says Google paid $2.7 billion in 2024 for him to return to work on cutting-edge projects, including Gemini, and that “Even that, though, couldn’t ensure that Shazeer got the computing resources he wanted.”
According to the WSJ, less than two years later he “sent a note to one of his teams saying that Google had cut what was allocated to them in favor of another project.” The paper adds: “Shortly thereafter, Shazeer left for rival OpenAI.”
Shazeer’s move to OpenAI was reported in June, and we covered it on 18 June 2026. What the WSJ adds is the compute cut that came shortly before he left; it does not name the project that received his team’s allocation.
Compute Brokered at the Top
The WSJ reports that executives elsewhere are also involving themselves in compute deals. It says Anthropic co-founder Tom Brown visited Elon Musk at xAI’s offices in Hawthorne, California, in late March to broker a deal to rent computing power, according to people familiar with the matter.
The paper recalls that Anthropic announced an agreement with Musk’s SpaceX in May to lease more than 300 megawatts of computing power, part of a deal it says was later revealed to be worth up to $45 billion.
It also reports that Anthropic CEO Dario Amodei spoke by phone with Meta chief AI officer Alexandr Wang earlier this year to try to source more computing capacity, according to a person familiar with the matter. Per the WSJ, Meta debated whether to give chips to Anthropic and “ultimately decided not to for now,” according to people familiar with the matter.
The WSJ cites SemiAnalysis for a 60% rise over the past year in the hourly cost of renting H100 computing power on a one-year contract; the supporting figures are the WSJ’s own.
What Compute Costs Now
The WSJ cites SemiAnalysis for a 60% rise over the past year in the hourly cost of renting computing power from Nvidia H100 chips on a one-year contract. It says Alphabet expects to make up to $200 billion in capital expenditures this year, much of it toward AI infrastructure.
The paper reports that cloud providers moved to require multiyear contracts with upfront costs “often as high as 30%,” sometimes with guarantors. Citing a person familiar with the accelerator, it says one Y Combinator startup last year needed 120 graphics processing units for a two-month training run but could not find a suitable contract, and that Y Combinator launched a cluster for its startups in July.
The Structural Read
The WSJ’s three Google items share one feature: each concerns who decides where compute goes, not how much Google has. The paper reports Alphabet expects up to $200 billion in capital expenditures this year, much of it toward AI infrastructure.
Outside Google, the paper’s examples move the same question to the top of other companies. It reports Tom Brown visiting xAI to broker a compute rental and Dario Amodei calling Meta’s Alexandr Wang, both attributed to sources familiar with the matter.
For smaller buyers, the WSJ describes the terms rather than the politics: multiyear contracts, upfront costs it says are often as high as 30%, and a 60% rise in one-year H100 rental prices that it attributes to SemiAnalysis.
Evan Conrad, chief executive officer of San Francisco Compute, as quoted by the WSJ (10 October 2026)
“Thereโs a big gold rush happening at the moment.”
Three Implications
GOOGLE’S RESEARCH TEAMS The WSJ reports that researchers are fiercely divided over which projects deserve compute and that Hassabis has expressed frustration about compute for long-term research, according to people familiar with the matter.
FRONTIER LABS BUYING CAPACITY The paper reports executives personally brokering compute: Brown at xAI in late March, and Amodei with Meta’s Wang, where it says Meta decided not to give chips to Anthropic for now.
STARTUPS RENTING COMPUTE The WSJ reports multiyear contracts with upfront costs often as high as 30%, sometimes with guarantors, and cites SemiAnalysis for a 60% rise in one-year H100 rental prices over the past year.
The analysis draws the line this way: “Training was a cost center: episodic, concentrated among roughly twenty frontier model labs, amortizable across trillions of tokens.”
As we read it, the WSJ’s Google account is about how compute is divided among uses inside one of those labs. The paper reports Hassabis’s frustration about compute for long-term research; it does not say which uses received more.
What Is Not Established
The WSJ attributes its accounts of Brin and Hassabis to people familiar with the matter; it states the Shazeer note and the divisions among researchers in its own voice. The syndicated text we read on 10 October 2026 carries no comment from Google, Brin, Hassabis or Shazeer, and no figure for how much compute any Google team was given or lost.
The WSJ does not say how often Brin reshuffles compute or what the formal allocation process is. Meta’s decision on Anthropic’s request is reported as made “for now.”
Business Engineer Framework
The State of The Inference Economy
A Business Engineer analysis of how AI moved from the training era to the inference era, and where value concentrates as a result.
The Wall Street Journal reported on 10 October 2026, citing people familiar with the matter, that Sergey Brin sometimes upends Google’s formal compute allocation process and that Demis Hassabis has expressed frustration about compute for long-term research. The paper also reports that Noam Shazeer told a team Google had cut its allocation shortly before he left for OpenAI.
A note on sourcing. This piece rests on the Wall Street Journal’s reporting, which we read in full on 10 October 2026 in the version syndicated by Mint; the paper cites people familiar with the matter for its accounts of Brin, Hassabis, Brown, Amodei and Meta, a person familiar with Y Combinator for the startup example, and SemiAnalysis for the H100 price figure. We have not independently confirmed those accounts, and the text we read carries no comment from Google. Nothing here is a forecast, and nothing here is financial or investment advice.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur, HackerNoon and Search Engine People. Find him on LinkedIn and Substack. See how we source.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
