LinkedIn, the professional social network with more than 1.3 billion users, has decided not to expand its data centers for the next fiscal year, according to a report by WIRED. The company plans to keep its investment in GPUs steady and its compute and storage footprint flat through fiscal year ending next June. This stands in sharp contrast to the aggressive data center buildouts by OpenAI, Meta, and Google.
A Prudent Approach to AI Infrastructure
Executives at LinkedIn say the decision to hold the line on infrastructure spending is driven by efficiency gains. Erran Berger, chief technology officer for engineering, stated, “One of the goals we've set is to try to basically keep our compute footprint flat or as close to flat as possible while shipping more compute-hungry things to production.” Raghu Hiremagalur, chief technology officer for infrastructure, emphasized the scale of the achievement: “I really want to double underscore that for a company of our scale, to say a full year we're going to do this with no incremental storage and compute is no small feat, but it's taken a ton of work to get there.”
The company was able to avoid big AI hardware spending because it found ways to use its existing GPUs twice as efficiently over the past six months. Berger noted that the efficiency gains could compound over time, enabling LinkedIn to get more out of future data center expansions when budgets eventually increase.
Optimizing the AI Pipeline
LinkedIn’s cost discipline comes after an earlier attempt to migrate to Microsoft Azure after Microsoft acquired the company in 2016. That move “didn’t make economic sense,” according to Hiremagalur, because Azure was growing rapidly and LinkedIn’s own demands were skyrocketing. In 2022, LinkedIn went all-in on its own data centers in Oregon, Texas, and Virginia. This ownership gave the company control over every detail of its technology, setting the stage for the current optimization push.
To keep compute flat, LinkedIn optimized every stage of the AI pipeline — from training models to serving user queries. Hiremagalur’s team developed measurement tools to track compute and storage usage by individual teams, then set up systems to enforce efficiency. The effort was driven by unsustainable cost trends: “Every query that's coming to our site has increased in cost over time,” Hiremagalur said, adding that data storage was doubling annually. “That is not a sustainable place to be.”
| Metric | Past Trend | Current Target |
|---|---|---|
| GPU efficiency | Baseline | 2x improvement in 6 months |
| Compute footprint | Growing | Flat for fiscal year |
| Data storage | Doubling annually | No incremental storage |
Implications for Enterprise IT
LinkedIn’s strategy is noteworthy for enterprise technology leaders facing pressure to invest in AI. Songyee Yoon, managing partner of Principal Venture Partners and a board member at server maker HP, commented, “It is encouraging for the industry. It suggests AI is beginning to move from experimentation into production discipline. The companies that win will not simply be the ones that spend the most on infrastructure.”
Berger and Hiremagalur believe that the new constraints will motivate engineering teams to get more creative when developing generative AI features. LinkedIn has already begun rolling out AI-based assistants that help users write messages, find jobs, and recruit candidates. By keeping infrastructure spending flat, LinkedIn is proving that significant AI capabilities can be delivered without a proportional increase in hardware — a lesson that could resonate across industries, including supply chain and logistics technology, where cost-efficient AI deployment is critical.
Sources:
