As we speak on the corporate’s Q3 2025 earnings name, the place Intel noticed its first revenue in practically two years due primarily to these lifelines, CEO Lip-Bu Tan and CFO David Zinsner defined how the corporate doesn’t but have sufficient chips. It’s presently seeing shortages that it expects to peak within the first quarter of subsequent yr — within the meantime, leaders say they’re going to prioritize AI server chips over some client processors because it offers with provide and demand.
“We anticipate CCG [Intel’s consumer chips] to be down modestly and DCAI [Intel’s server chips] to be up strongly as we prioritize capability for server shipments over entry stage consumer components,” Intel says. Tan revealed immediately that Intel may even launch new AI GPUs every yr, following Nvidia and AMD in shaking up their conventional cadence to handle the large demand for AI servers. It’s not clear what that may imply for these hoping for extra Intel gaming GPUs.
Whereas all eyes are on Intel’s sizzling new Panther Lake and its 18A course of to point out the world it may nonetheless take advantage of potent client PC chips and make them in-house, the corporate reiterated it’s solely launching one SKU of Panther Lake this yr and slowly rolling out others in 2026. Right here’s one other attainable motive why: Zinsner hinted immediately that Panther Lake can be a “fairly costly” product to start out with, and Intel’s going to should push its current Lunar Lake chips as a substitute “in at the least the primary half of the yr.”
Whereas Intel has repeatedly pushed again towards the concept its 18A course of had poor yields, the corporate admitted to buyers and analysts immediately that it’s not able to be an enormous monetary success both: yields are “ample to handle the availability however not the place we’d like them to be to drive the suitable stage of margins,” says Zinsner, suggesting that it may be 2026, and even 2027 for an “acceptable stage of yields” in that regard.
For now, Intel can be “working intently with clients to maximise our obtainable output, together with adjusting pricing and blend, to shift demand in the direction of merchandise the place now we have provide they usually have demand” — which seems like taking part in with the costs it prices PC makers to stay Intel inside their computer systems and pointing them at Lunar Lake components as a substitute of sizzling new ones. Tan reiterated immediately that he’s not going to put money into extra capability until there’s “dedicated exterior demand,” and Zinsner says investments in capability subsequent yr received’t “considerably change expectations”.
Intel says that 18A can be a “long-lived node” that may energy “at the least the following three generations of consumer and server merchandise.” If you happen to had been hoping for a return to the “tick-tock” days the place Intel would alternate between shrinking its chips and releasing new architectures each technology, that’s not taking place right here.
However that doesn’t imply Intel will cancel its subsequent node, Intel 14A, because it warned it’d. Tan prompt immediately that clients have stepped in to avoid wasting 14A, and Intel, that the corporate is “delighted and extra assured” in it, and Zinsner says it’s not solely “off to a superb begin,” however higher than 18A was at this level “by way of efficiency and yields.”
