A global shortage of raw computing power is upending Silicon Valley, sparking bitter corporate turf wars, sending executives on worldwide scrambles for hardware, and forging improbable alliances between fierce rivals, according to a report by The Wall Street Journal (WSJ). The fierce bottleneck has become the tech industry’s central battleground, reshaping executive suites and imperiling the prospects of up-and-coming startups that cannot secure enough chips to keep pace.
The severity of the compute crunch was laid bare inside Google, which spent $2.7 billion in 2024 to rehire pioneering researcher Noam Shazeer to steer key projects like its flagship Gemini models, according to a WSJ report. The Journal reported that Google co-founder Sergey Brin has at times overridden formal allocation procedures to divert compute to his preferred projects, while DeepMind co-founder Demis Hassabis has expressed frustration over a lack of resources dedicated to fundamental, long-term research.
“One of the biggest points of friction inside Google is compute, according to an earlier report in CNBC. ” Claiming sources it further said that frustrations over access to compute can be especially acute when Google announces large infrastructure commitments to competing labs like Anthropic, whose models compete directly with Gemini. “Anthropic CEO Dario Amodei also spoke on the phone with Meta chief AI officer Alexandr Wang earlier this year, attempting to source more computing capacity, according to a person familiar with the matter, according to the report. Meta debated whether to give chips to Anthropic and ultimately decided not to for now, people familiar with the matter said.
Despite the enormous price tag, Google failed to guarantee Shazeer the computing capacity he sought. After internal reallocations cut his team’s resources in favor of another initiative, Shazeer departed for rival OpenAI, highlighting an internal environment where researchers frequently quarrel over processing power. The report added, “Google is investing more than almost any company in the world in data centers, chips and related infrastructure. But capacity remains scarce. Every TPU assigned to training a model, serving a Google product, or fulfilling a contract with a cloud customer reflects a choice among competing priorities. Amodei also reportedly approached Meta’s chief AI officer Alexandr Wang in search of excess capacity. “