Intel's 18A Push and the 14A Bet
Intel's got itself in a spot, and CEO Lip-Bu Tan isn't exactly hiding it. During the company's Q2 2026 earnings call, Tan laid out a candid assessment of where things stand: Intel needs to catch up to AMD and Arm, and it's going to take everything the company has—along with a few process nodes it hasn't shipped yet—to make that happen.
Revenue came in at $16.1 billion, up 25 percent year-over-year, which is solid. That's the seventh straight quarter of beating financial guidance, a streak that would impress almost anyone. But here's the thing that keeps data center operators up at night: Intel posted a GAAP loss of $11 billion for the quarter. Non-GAAP losses clocked in at $2.2 billion. The revenue looks good on paper. The bottom line? Not so much.
"I'm optimistic about Intel's prospects," Tan said, which is executive-speak for "we're spending a fortune and we think it'll pay off eventually." The optimism stems from the AI boom driving demand across multiple product lines, plus Intel's foundry service, which the company is betting will become a standalone growth engine rather than just supporting internal needs.
"Our core server CPU franchise is growing faster than ever," Tan told investors. He pointed to the Xeon 6 range as "one of the fastest-ramping products in Intel's history." That's encouraging, sure. But it's also the kind of statement you make when you're trying to convince people the ship isn't sinking quite as fast as it might.
The Server Market: Arm Takes Half, AMD Takes a Third
Here's the uncomfortable math. Intel's hyperscale customers—the cloud operators building massive data centers—have been designing their own Arm-powered CPUs. They're deploying them in such numbers that analyst firm IDC found non-x86 servers now account for almost half of all server sales. Half.
AMD, meanwhile, has grown its market share to one-third of the x86 server market. That leaves Intel with roughly a sixth of the total addressable space, which is... not where you want to be when you're the incumbent.
Morgan Stanley analyst Joe Moore pressed Tan directly on the earnings call about regaining market share. The CEO's answer? Point to upcoming products and hope the market agrees.
Tan highlighted Intel's forthcoming Clearwater Forest, Diamond Rapids, and Coral Rapids processors as evidence the company is building products competitive with anyone's. Clearwater Forest is the next-generation server CPU. Diamond Rapids targets networking and storage workloads. Coral Rapids is the memory-optimized variant for analytics. Together, they represent Intel's best shot at relevance in a market shifting beneath its feet.
"Some areas we are still behind," Tan admitted, which is about as close to "we're losing" as a CEO will get on a public call. "But we are catching up very fast and we try to leapfrog some of the CPU architecture, and we are putting major effort into it. Time will tell."
That last phrase—"time will tell"—carries a lot of weight. It's the sound of a company betting its future on process nodes that aren't in volume production yet.
Supply Chain Headwinds: Tighter Than Ever
If you thought the semiconductor industry was operating at full capacity, think again. Tan described "one of the most severe supply constraints in its history" across leading-edge logic silicon wafers, memory, and substrates. These shortages, he said, "will persist for the foreseeable future."
That's not a near-term problem. That's a multi-year constraint that affects everyone from Intel to AMD to Arm-based chip designers. It also means Intel's foundry ambitions face a bottleneck that isn't entirely within its control.
There's at least one silver lining: wafer output across Intel's major manufacturing nodes exceeded expectations from 90 days prior to the earnings call. That's a positive signal, even if it doesn't change the fundamental supply situation.
"Supply remains very tight and the near-term linearity of our supply growth is more skewed towards the end of Q3 and into Q4, especially for servers," Tan said. Translation: if you're ordering server CPUs today, you might wait. A lot.
The 18A and 14A Process Roadmap
Intel's foundry strategy hinges on two process nodes: 18A, which is already in production, and 14A, which is still 18 months away from risk production for internal products.
The 18A process, Intel's current flagship, is now in volume production across multiple commercial and consumer products for the PC client segment. Tan reported that "factory output continues to increase sequentially every month" and that yields from 18A are "trending ahead of targets." Those are encouraging metrics, and the successful high-volume ramp of 18A for internal products provides what Intel calls "important validations" as the company courts external foundry customers.
Then there's 14A, the node Intel hopes will make it a more formidable competitor to TSMC. Tan laid out the timeline clearly: risk production for internal products in the second half of 2027, with a full commitment to high-volume ramp in 2028.
That's a long runway. It's also a commitment that signals Intel's foundry business is moving from supporting role to primary growth driver. Whether the market will embrace Intel as a foundry partner remains to be seen. TSMC has a significant lead, and the supply constraints Tan described affect everyone equally—including Intel's competitors.
The Edge AI Pivot: From PCs to Physical AI
Here's where things get interesting. Intel recently renamed its PC business the "Client Computing and Physical AI Group"—or CCPG, if you prefer the acronym. That's not a cosmetic change. It reflects a strategic bet that edge AI and physical robotics could become as valuable as the traditional PC business.
CFO David Zinsner told investors that "the edge and physical AI opportunity is likely to at least match the client TAM [total addressable market] over time." That would make edge and physical AI an roughly $8 billion business, given that edge products represented ten percent of CCPG's $8.9 billion revenue in the quarter.
CCPG revenue rose 13 percent year-over-year, which Zinsner called "pleasantly surprising." The PC market itself, however, is "softer" due to "memory dynamics in the marketplace," and Zinsner predicted Q3 performance would fall.
Tan added that Intel "still has work to do to establish a strong footprint in the edge and physical AI ecosystem but see this opportunity as an important future growth driver."
That's a cautious formulation. It's also a recognition that edge AI isn't the silver bullet it's often painted as. Building a strong ecosystem around edge devices and physical robotics takes time, capital, and partnerships that Intel is still assembling.
What This Means for the Market
Intel's Q2 2026 results paint a picture of a company in transition. Revenue is growing. Losses are massive. The server market is shifting toward Arm. AMD is gaining share. Supply constraints are severe. And Intel is betting its future on process nodes that won't reach high-volume production until 2028.
Tan's candid admission that Intel needs to "leapfrog" AMD and Arm in CPU architecture isn't just a press release talking point. It's a recognition of where the market stands today—and a roadmap of where Intel hopes to be in two years.
The 18A process is performing ahead of targets. The 14A roadmap is on schedule. The edge AI opportunity is real, even if it's still maturing. Whether all of this adds up to a competitive Intel remains the question.
"Time will tell," Tan said. And in the semiconductor business, time is both the enemy and the ally.