It is no secret that there is a considerable shortage of storage devices, both traditional spinning media and SSDs. It’s always been hard to quantify beyond the “they are out of stock” complaints, but now some companies are getting tangible stats.
Large-capacity hard drives, measuring 18TB or more, now cost up to three times more than their typical prices due to production capacities being entirely booked by hyperscalers. Among those feeling the pinch of the shortage are archivists like The Internet Archive, which runs the Wayback Machine.
“We have found that the preferred 28-30TB drives are just not available or at very high price,” Founder Brewster Kahle told 404 Media. “We gather over 100 terabytes of new materials each day, and we have over 210 Petabytes of materials already archived on machines that need continuous upgrades and maintenance, so we need to constantly get new hard drives.”
The Wikimedia Foundation, which runs Wikipedia and a host of related projects, told 404 Media roughly the same thing. “We see the primary impact in the purchase of memory and hard drives but also in terms of lead times on server deliveries and our capacity to place future orders,” a Wikimedia Foundation spokesperson told them.
Gartner believes there will be a shortage in traditional hard drive as well as SSDs through at least the first half of 2027. “Things might loosen up a little bit in Q4 of 2027, but the storage demands are so large because hyperscalers for AI are commandeering all of the storage because they’re willing to pay for it, and that is squeezing everyone else in the world,” said Joseph Unsworth, senior research analyst with Gartner.
In the seller’s market, the advantage will always go to the customer with the deepest pockets and willing to make the largest purchase. That inherently favors hyperscalers, who buy in mass quantities versus enterprises and SMBs, Unsworth said.
Hard drives have seen their use fade as flash memory cheapens and SSD storage capacity grows, but they are still widely used in the data lakes that constitute AI models. “Hard drives are used for the majority of the data lakes. That is putting pressure on every other customer out there. In fact, even tape demand has surged, and prices have gone up and lead times have extended,” Unsworth said.
This leaves enterprises and medium sized businesses that may need a storage array of around 100-200TB or so, modest by comparison to a hyperscaler, in a difficult position.
“We’re basically saying, analyze your workloads to make sure you’re using the right technology for the right workload, said Unsworth. “And if you don’t need high performance, you shouldn’t be paying the flash premium. You can use hybrid arrays. Hard drive-based arrays can be good enough for those workloads, especially if your budget didn’t go up. And I can’t imagine many corporations’ budgets will go up.”
The other option to consider is extending the life of existing storage arrays. Storage arrays are built to last much longer and rather than replace them after three years, they can go five to seven years, said Unsworth.
“This is not the time to refresh your storage array this year, or I would even argue, even next year. So if you can extend that, as long as you have the right maintenance program in place, and you’re able to get access to software updates and security patches, we think that’s a very viable solution for a lot of customers,” he said.