Intel Digs Deep Inside to Make Foundry Push Work

Both were readily apparent on Wednesday, when the chipmaker’s stock fell after it hosted an investor call to discuss its “internal foundry model.” No major new news was given. Instead, the event essentially elaborated on an announcement the company made in October about how its business units would operate as it seeks more business manufacturing chips designed by other companies. But some investors seemed to be hoping for an announcement of the first major customer for that business. Intel’s stock fell 6% by the closing bell—more than double the selloff that the PHLX Semiconductor Index suffered during the rough session.

Intel indeed has a lot riding on its ability to eventually land big customers for its foundry services segment, also called IFS. The Silicon Valley pioneer has spent the bulk of its 54-year history manufacturing chips of its own design. Opening its factories to other chip companies is no simple process—especially because many of those potential customers make chips that compete with Intel’s. The IFS segment was actually first announced by Intel more than two years ago, during a time of severe shortages globally of chip-manufacturing capacity. The segment still generated only $730 million in revenue for the trailing 12-month period ended in March—1% of the company’s total revenue for that period.

As Intel made clear on Wednesday, making more chips for outsiders requires some major changes on the inside. The so-called “internal foundry model” essentially makes Intel’s PC, data center, automotive and other chip design units into customers of its own factories. That means the managers of those units bear responsibility for profits and losses, making them accountable for things such as accelerated production orders, or “expedites.” Intel said Wednesday that its fabs process such “expedites” about 2 to 3 times more often than peers, which has caused an 8% to 10% hit to its overall output. If Intel’s internal customers have to pay up for those expedited orders, they might think twice.

Reducing such inefficiencies is key to Intel’s ability to deliver on its aggressive cost-cutting goals. The company said Wednesday that cutting back the aforementioned expedites alone is expected to save as much as $1 billion a year over time. And creating some internal separation between the product groups and the manufacturing business could also increase the chances of landing major customers who might have been nervous about sharing their chip designs with a competitor. Mark Lipacis of Jefferies called the changes “a step in the right direction,” and raised his price target on Intel’s stock by 6% in a note Thursday morning. Ruben Roy of Stifel called the changes a “significant business transformation” in his own report Thursday.

But to land the most desirable customers, Intel still needs to catch up its manufacturing process with those of foundry leaders Taiwan Semiconductor Manufacturing, or TSMC, and Samsung. The Wall Street Journal reported last month that Tesla and Qualcomm both recently passed on signing deals with Intel’s foundry business over concerns about its most advanced production processes and other technical issues. Intel is working on closing that gap, and said it expects to announce a major foundry customer for the most advanced process it is developing before the end of the year.

Morgan Stanley analyst Joe Moore says “a major customer win would be a positive for the story.” But in a report Wednesday, he also noted the risk that the changes required to build up a foundry business could also cost Intel some competitive edge in its core business of designing and producing CPU processor chips for products such as PCs. Intel’s share price has jumped 24% this year on signs that its PC business has bottomed, but the stock still lags behind most of its chip peers, primarily on worries that its expensive overhaul won’t eventually bear fruit. Intel still has a lot of work ahead.

Write to Dan Gallagher at [email protected]