AWS says acute server memory shortage is driving customers to the cloud
When you can't get 'em with a 'transformation plan,' supply chain pain will do the job
The great memory shortage is having yet another effect, pushing enterprises into the waiting arms of the cloud operators as they can't secure enough on-prem compute themselves.
Or so claims AWS chief Andy Jassy, who says the memory supply situation is forcing many firms to bring forward plans they had to move resources to the cloud.
"On memory and storage and the supply chain, I think everybody knows that the cost of components, particularly memory, has skyrocketed. We are in a stage where there is just not enough capacity for the amount of demand," Jassy said during an analyst call for the company's first quarter earnings on Wednesday.
"One of the interesting things that we see right now with the change in price and supply on things like memory is that it is a further impetus pushing companies who have on-premises infrastructure into the cloud," he added.
"We have seen a number of conversations we have been having with enterprises for many months - where it has just been slower in getting the transformation plan to move to the cloud - accelerate rapidly just because we have a lot more supply than what others have. It will be interesting to see how that evolves over time."
AI is the root cause of the memory chip shortage. Due to insatiable demand for high-performance and costly DRAM and HBM for AI infrastructure, chipmakers have switched manufacturing capacity to churn out these rather than the more standard memories needed for other applications.
As a massive multi-national company - AWS generated $37.6 billion in revenue during the quarter – the cloud biz is at the front of the queue when it comes to obtaining the hardware to keep expanding, as are the other hyperscale operators.
"A meaningful part of those suppliers are prioritizing their very largest customers, which cloud providers are," Jassy noted, but implied that some planning ahead on his company's part played a role.
"We saw this trend happening early, in the middle to latter part of last year, and we have worked with our strategic suppliers to get a significant amount of supply," the AWS supremo explained.
"I think we have done a good job in making sure that we are not capacity constrained there, but we will watch that very closely."
If this trend is true, it would be a reversal of the situation just 18 months ago, when the cloud giant claimed it was facing stiff competition from on-premises infrastructure, in evidence given to the UK's markets watchdog, the Competition and Markets Authority (CMA).
AWS isn't the only company making the claim. A senior exec at IT solutions firm Insight Enterprises told CRN that he is seeing an acceleration of clients exiting their datacenters and going to Google Cloud due to several reasons, including a lack of access to hardware.
"There's a couple of things driving it: chip shortages, less access to the gear, better access to the AI models in [the cloud]," said general manager and senior VP Peter FitzGibbon.
But Omdia Chief Analyst Roy Illsley cast doubt on this, saying that while some smaller organizations may struggle to get the kit they need, he doesn't believe it is a trigger for a wave of cloud migrations. Instead, it is a bit of wishful thinking on the part of cloud operators like AWS, spreading a bit of FUD (fear, uncertainty, and doubt).
"Cloud is a strategic choice that customers make, and most on-premises datacenters already have servers with memory, so it just means they delay a refresh or upgrade and wait for a server to be delivered," Illsley told The Register.
It then comes down where you are in the priority list to get a server, he added, and hyperscalers are top of that list, "But do they have the capital to buy all the servers? No, they do not. So, the OEMs will get some and these will flow to enterprises. Yes, smaller enterprises may struggle and switch to cloud, but it is no way a flood," he stated.
Gartner VP analyst for Infrastructure and Operations Tony Harvey confirms it is happening. "Yes, we are seeing a lot more interest in cloud due to the increase in server prices and delays in shipment, as well as vendors shortening quote times and not guaranteeing pricing until shipment," he told The Register.
"From what I see it's a combination of availability, you can get it now, and a different cost trade-off. When an on-premises server costs 4x what it did a year ago that changes the comparison to cloud, and so far, the major cloud vendors have not increased prices on non-GPU servers."
Elsewhere, Meta is reportedly extending the life of its servers as it grapples with a "significant server supply deficit" due to the shortage of memory chips.
According to the Wall Street Journal, the server life cycle is being extended from six to seven years, as the social media giant did not anticipate the hardware demand growth now seen in the industry.
Supplies of critical server components including DRAM and hard drives are expected to be constrained through 2027, Meta believes.
As The Register has previously reported, hard drive manufacturers have already sold all the units they will make this year, while there is expected to be shortage of some server CPUs that could see prices increase by up to 15 percent. ®