Big Cloud is poised to corner the market for enterprise hardware

Hyperscale cloud providers, including Amazon Web Services, Meta, and Microsoft, are leveraging their financial muscle to secure scarce enterprise hardware, potentially cornering the market and forcing businesses to rent capacity. Nutanix CEO Rajiv Ramaswami noted in May that renting from a hyperscaler is now faster than waiting for hardware delivery, while Amazon CEO Andy Jassy said AWS recoups server costs in under three years and expects the business to become a trillion-dollar annual revenue stream. Memory makers Micron and SK Hynix and hard disk maker Seagate have prioritized big customers, and Meta CEO Mark Zuckerberg said the company is receiving offers for compute at a significant premium.

Big Cloud is poised to corner the market for enterprise hardware Source: The Register https://www.theregister.com AI gave hyperscalers first dibs on scarce kit, and business buyers may have little choice but to rent it back OPINION Whenever I talk to execs from hyperscale clouds, they suggest it is their destiny to eventually host the majority of enterprise workloads. That future is now in sight because the AI boom has shown how they can use their financial muscle to command the hardware supply chain. Nutanix CEO Rajiv Ramaswami made the situation plain in May when he pointed out that the fastest way to access a new server was to rent it from a hyperscaler rather than wait for an established hardware provider to deliver one. Ramaswami's observation reflected hyperscalers' superior buying power, made possible by the fact that component suppliers happily put them at the front of their queues for kit in the current supply-constrained environment. Evidence of that preference is not hard to find. Memory makers Micron and SK Hynix recently went on the record about long-term deals that guarantee supplies to their biggest customers while all but locking in historically high margins. Hard disk maker Seagate has done likewise. AMD, meanwhile, has sweetheart deals with OpenAI /glossary/openai and Meta. Speaking of Meta, CEO Mark Zuckerberg recently used the company's earnings call to argue that The Social Network's balance sheet means it can readily attract investment to fund capital expenditure. Discussing Meta's ambition to start an infrastructure-as-a-service business, Zuckerberg said the company is already "getting a lot of offers for compute /glossary/compute at a significant premium over what we paid for it." That's an important observation because it suggests Meta doesn't pay top dollar for hardware. Once Meta is in the business of renting AI infrastructure, it will have the tools it needs to rent conventional compute and storage too. Amazon CEO Andy Jassy has also explained how the economics work, telling investors that Amazon Web Services recoups its spending on servers and networking equipment in under three years. "The servers currently have a useful life of at least five to six years, and most of our AI capacity these days is being contracted for at least five-year terms," he said. "That means that we're driving significant free cash flow on the servers… after we break even." Jassy expects those economics to improve as AWS shortens its breakeven times, allowing servers to pay for themselves more quickly. The CEO also said Amazon expects its datacenters to last 30 years, with margins on the hardware inside them rising further once the buildings have paid for themselves. Amazon says it currently ties infrastructure spending to predicted demand. But if hyperscalers secure most of the available hardware, more organizations may have little choice but to rent capacity from them, bringing that demand with them. Major server makers like Dell, HPE, Lenovo, and Supermicro retain one excellent argument for buying rather than renting: fewer nasty surprises on the bill, albeit at the cost of less elastic infrastructure. Yet big clouds don't see that as a threat. Indeed, they see years of growth ahead. "We long believed AWS could become a few hundred billion-dollar revenue business and now believe it'll be at least double that, and very possibly be a trillion-dollar annual revenue business for us in time," Jassy told investors on Amazon's last earnings call. Smaller clouds compete by undercutting majors like Amazon. They, too, appear to be buffeted by supply chain problems, as shown by OVH's steep price rises while its larger rivals have yet to increase customer charges. Hyperscalers' suppliers won't mind this because it means they can focus on a small collection of big customers, which reduces their sales and marketing costs and preserves their margins. Meanwhile, no current enterprise hardware vendor has a path to becoming a trillion-dollar company. Only Dell has cracked $100 billion. This leaves organizations that prefer owning infrastructure to renting it from clouds in a tricky position: waiting months for hardware and hoping vendors will honor quotes. And all this woe was made possible by AI. ®Get AI news in your inbox Daily digest of what matters in AI.