LinkedIn Bucks AI Build-Out, Holds Data Center Spend Flat
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- LinkedIn will keep GPU investment steady and its compute and storage footprint flat for the fiscal year that began last month and ends next June, executives told WIRED — making it the largest business yet to publicly defy the AI building boom.
- CTO Erran Berger said LinkedIn roughly doubled GPU efficiency over the past six months through pipeline-wide optimization, including model distillation for job recommendation models, workload balancing between CPUs and Nvidia GPUs, and rewriting foundational software on Nvidia chips to handle larger tasks.
- LinkedIn estimates its efficiency work saved about $24 million over the past 12 months — equivalent to roughly 1,100 GPUs running around the clock for a year — though the executives acknowledged that's modest against the platform's $18 billion in annual sales.
- CTO for infrastructure Raghu Hiremagalur said server prices have jumped three-fold in recent months, but LinkedIn locked in some savings by buying ahead and will still purchase new machines to replace aging or broken hardware without expanding its footprint.
- LinkedIn went all-in on its own data centers in Oregon, Texas, and Virginia in 2022 after deciding it didn't make economic sense to squeeze onto parent Microsoft Azure's general-purpose infrastructure, and reports north of 95 percent GPU utilization on the training side.
- Gartner analyst Chirag Dekate framed the shift as a move from a "buy-more era" to a "do-more era" — part of a trend sometimes called "tokenomics" — but warned that firms tightening too much may eventually have to compromise on AI ambitions.
- Principal Venture Partners managing partner Songyee Yoon, who also sits on HP's board, called LinkedIn's stance "encouraging for the industry," arguing that "the companies that win will not simply be the ones that spend the most on infrastructure."
Why it matters: For a company with $18 billion in annual sales, $24 million in savings is rounding error — but the playbook matters more than the number. LinkedIn's ability to double GPU efficiency through software rewrites gives enterprise CFOs a concrete counter-narrative to capex-heavy peers like Meta and Google, and its public stance makes "buy more to save more" a harder assumption to defend in boardrooms.
