Major chip vendors including Intel are telling customers that they’re set to begin hiking chip prices in the fourth quarter as they deal with increasing costs and supply shortages.
The rumor initially started with Intel. Now Qualcomm and Marvell Technologies have also begun informing customers of price increases, according to a report from Taiwanese publication DigiTimes.
“On its Q1 earnings call, Intel indicated it would increase pricing in certain segments of its business due to inflationary pressures. The company has begun to inform customers of these changes,” an Intel spokesperson said in a statement to me.
Just how high remains unclear. News publication Nikkei Asia cited sources who claim the increase will be as much as 20%. But Intel did not comment on the rate of increase.
AMD has not spoken on the issue, and when reached for comment, declined, citing its “quiet period” between the end of the fiscal quarter (June 30 for AMD) and when it will announce earnings (August 2). But AMD’s chips are made by TSMC, and TSMC is also alerting customers to price hikes.
The fact that it’s Intel means that the problem isn’t necessarily one of capacity. Intel has its own manufacturing capacity and can quite adequately meet it. The real problem is resource materials. CPUs are more than just silicon; they are a combination of very rare, hard-to-formulate chemical compounds which have been in short supply for some time.
Jim McGregor, principal analyst with Tirias Research, said he isn’t surprised at this news. “A year ago, foundries told customers [they] need to put money into capacity or face lower priority and/or increasing prices,” he said.
The increase is expected to be across-the-board, from low-end to high-end processors. Server processors will really feel the pinch of that 20% increase if it is realized. Xeon Scalable processors range from $2,000 to $8,000. Imagine a 20% price hike per chip. Now multiply that by two, since most servers are dual socket, and then multiply that by however many hundreds of servers are being purchased at one time.
That’s why this will hit cloud service providers particularly hard, since they tend to buy servers in the tens of thousands. But whether it’s AWS or your own internal deployment, you can best believe that the server vendors will pass on the costs to you.
The one potentially redeeming factor, according to McGregor, is that all the large server OEMs have deep contracts with Intel and other chipmakers. While chip makers often have listed prices, those do not apply to tier 1 OEMs because they have the volume deals. But that doesn’t mean there will be zero impact, just less of an impact for HPE and the like, while smaller OEMs will likely take a bigger hit.
McGregor added he believes the shortage is caused by a multitude of events, from labor to shipping to shortages of key supplies to more recent shutdowns in China. “There are constraints throughout the supply and value chain, and there’s nothing you can do about it,” he said.