Hard drive manufacturers have already sold all the units they will make this year, and it looks like the AI infrastructure boom is to blame, with hyperscalers soaking up all the high-capacity storage.

Seagate and Western Digital, two of the Big Three makers of rotating disk drives, confirmed during recent earnings conference calls that they have already flogged all of their manufacturing output.

Western Digital chief Tiang Yew Tan told analysts “We’re pretty much sold out for calendar ’26. We have firm purchase orders with our top seven customers. And we’ve also established long term agreements with two of them for calendar year ’27 and one of them for calendar year ’28.”

Seagate CEO Dave Mosley said on his own company’s call: “Our nearline capacity is fully allocated through calendar year 2026, and we expect to begin accepting orders for the first half of calendar year 2027 in the coming months… additionally, multiple cloud customers are discussing their demand growth projections for calendar 2028, underscoring that supply assurance remains their highest priority.”

Great for WD and Seagate. Not so much for eveveryone else.

It is understood that the third major disk producer, Toshiba, is likely to be in a similar situation. We asked the drive maker and will update if we get an answer.

Hard drives have pretty much been displaced from everyday PCs and laptops, but the makers continue to bump up their capacities, making them attractive for applications requiring large volumes of data storage at low cost – such as cloud storage and AI training. Western Digital recently announced it will ship 44 TB drives this year, for example, with a roadmap to 100 TB by 2029.

What are the implications for this impending shortage? “In my view this means no meaningful open production remains for discretionary buyers except the hyperscalers,” said Sid Nag, President and Chief Research Officer at Tekonyx.

“I am guessing these include locked-in purchase agreements extending into 2027 and 2028 for certain customers including those building AI datacenters. So there should be no impact to that business. This also implies HDD manufacturing capacity is now almost exclusively prioritized for large AI/cloud players because of predictable, high-volume demand,” Nag told us.

“This will, however, impact the mid-size market who depend on server technology.”

Omdia Senior Research Director for Enterprise Infrastructure Vlad Galabov agreed.

“I think it will be a very tough year for the standard enterprise general purpose server and for enterprise storage. We have downgraded our forecast for those markets,” Galabov told The Register.

Despite this, Omdia raised its overall 2026 server spend forecast to $590 billion and its datacenter capex forecast to more than $1 trillion because the top ten cloud providers keep upping their investment plans. 

Corporate IT projects will likely take a hit if they were counting on using hard drives as a capacity tier in their storage plans, but there are already shortages of DRAM and NAND flash silicon, the latter used in solid state drives (SSDs) – all due to AI-driven demand putting a strain on supply chains.

“We are seeing shortages of memory, storage and even CPU silicon and all of these will be dynamically affecting each other for some significant time based on announced capex spend and (datacenter) land leases,” said IDC’s senior research director for European Enterprise Infrastructure, Andrew Buss.

“The growth of AI demands storage, and lots of it, and also networking to move the stored data around, so expect there to also be high-performance networking shortages as well going forwards. And with next-generation Rubin GPUs apparently needing 20TB+ of fast SSD storage capacity per GPU, this will become even more acute,” Buss told us.

“This has been consuming large amounts of fast flash-based NVMe SSDs, and the increase in price and shortage of media is now encouraging many who need storage to reconsider HDDs. We would expect hybrid flash arrays to have a resurgence, but also HDD only arrays where the traffic patterns are suitable,” he added.

So it looks like it might be a bad year for a server refresh, unless you happen to be one of the hyperscalers, with shortages and price hikes affecting many components.

Omdia forecasts the top ten cloud service providers (Google, Amazon, Microsoft, Meta, Oracle, CoreWeave, ByteDance, xAI, Alibaba, and Tencent) to account for more than 70 percent of the server capex this year and AI-optimized servers for 80 percent of the total server spend. ®