Anthropic Ramps Up Hiring Spree to Tackle AI Infrastructure

The AI industry’s insatiable appetite for computing power is reshaping corporate recruitment strategies, with Anthropic’s latest hire underscoring how AI infrastructure demands are driving a high-stakes talent war among technology giants.

Anthropic has appointed Sana Ouji, a senior energy and data centre executive from Google, as the San Francisco-based AI company accelerates efforts to build a substantial global infrastructure portfolio.

Sana spent more than six years at Google, most recently focusing on data centre energy strategic investments and partnerships. Her move represents the latest example of how AI firms are prioritising infrastructure expertise as computing capacity becomes increasingly critical to competitive advantage.

The hire could signal a broader shift in how AI companies are approaching their operational foundations. Rather than treating data centre management as a secondary concern, leading developers appear to be elevating infrastructure to a strategic priority on par with algorithm development and model training.

AI compute demands reshape talent priorities

Sana’s departure marks the latest in a string of senior infrastructure appointments Anthropic has made from Google, reflecting how talent shortages are forcing AI companies to aggressively pursue experienced executives.

Announcing her move on LinkedIn, Sana framed the decision as a deliberate step towards greater challenge rather than straightforward career progression.

“After 6.5 years at Google, I'm taking on a new challenge,” she says. “The past few years in particular have been transformational for the energy and data centre industry, and to have had a front row seat to that transition at one of the world's leading AI companies has been a true privilege.”

She joins Ariel Horowitz and Tim Hughes on what Anthropic is calling its inaugural energy team – a designation that speaks to how recently the company has begun formalising this function.

Ariel, who joined in March, previously served as deputy director of grid modernisation for the US Department of Energy before the department was significantly restructured in April 2025. Tim, meanwhile, joined in February, arriving from data centre firm Stack Infrastructure, where he had been chief development officer.

Google exodus reveals infrastructure priorities

Sana is far from alone in making the crossing from Google to Anthropic. The company’s head of data centre infrastructure, Winnie Leung, is a former Google executive – as is Brett Rogers, who previously led data centre construction at the firm.

Other appointments from Google include Liwen Mao, now Anthropic’s data centre design lead, Adam Johnson, the firm’s data centre electrical lead, and Peter Sarossy, a Google veteran of 20 years, who joined in January as a data centre security engineering staff member.

Zach Miller, after 17 years at Google, became Anthropic’s data centre operations manager, and Soheil Farshchian, formerly a data centre system architecture lead at Google, is also believed to have joined.

Anthropic is, in effect, siphoning off a significant volume of Google’s infrastructure institutional knowledge into its own organisation. This systematic recruitment could suggest that established technology companies’ experience in managing large-scale computing infrastructure has become a crucial asset for emerging AI developers attempting to scale rapidly.

The compute arms race

In October 2024, Anthropic signed a deal with Google for cloud access exceeding 1 GW, including up to one million of Google’s tensor processing units (TPUs).

In April 2025, that relationship expanded further, with Anthropic agreeing terms with Broadcom and Google for the supply of TPUs representing 3.5 GW of capacity.

Separately, the company has pledged to invest $50bn in US data centres through a partnership with Fluidstack, with Google providing financial backing for those projects. These figures illustrate the scale of investment AI companies believe is necessary to remain competitive in model development.

Rival developer OpenAI has not been shy about exploiting the optics of Anthropic’s infrastructure position, claiming in an internal memo that Anthropic made a “strategic misstep to not acquire enough compute”.

Anthropic’s CEO Dario Amodei has argued that the calculus is genuinely difficult, warning in February that being even a year out on growth projections, or misjudging the rate of expansion, could be enough to “go bankrupt”.

Against that backdrop, the assembly of a dedicated global energy team – and the calibre of names being recruited to it – signals that Anthropic is treating infrastructure not as a back-office function, but as a strategic priority in its own right.

As AI models grow more sophisticated and computationally intensive, the companies that secure reliable, scalable computing infrastructure may gain decisive advantages over competitors still scrambling for capacity.