Intel Q425: Back to Reality

Continued supply constraints, advanced packaging timing, and why Q1 2026 will be more of the same

From WSJ’s How Intel Came Crashing Back To Earth After It’s Trump Bump:

When President Trump promised Intel nearly $9 billion and gave it his vote of confidence as an America-first tech company, it looked like the start of a new era. Investors assumed new orders would flow to the troubled chip maker and bid up the stock 120% in just five months.

And customer demand for Intel’s products did explode—but Intel wasn’t ready for it. After months of cutting capacity on its older production lines, the company was unprepared for a surge of orders for processors to put in AI data centers. Intel’s stock has crashed 17%, wiping out more than $46 billion in market value, since executives revealed the flub on the company’s fourth-quarter earnings call Thursday.

“The stock went vertical on vibes and tweets,” said Stacy Rasgon, a semiconductor analyst at Bernstein. “In theory, they should be in place to capitalize on this demand, but they’re not. What a shame.”

Stacy is right. The Intel stock run-up into Q4 earnings was built on tweets and vibes. Irrational, tbh. But the correction afterward shouldn’t have been a surprise to anyone. Honestly, it was a return to reality.

Yes, rumors of foundry customers are a positive sign. But customers wouldn’t be announced on this call. That’s just not what LBT does. Furthermore, 14A design decisions happen in 2H26 to 1H27. Which doesn’t even guarantee we’ll hear announcements in that time frame; even if decisions are made in that window doesn’t mean the customer wants it announced immediately. Customer announcements are a lagging indicator.

Moreover, the market treated Intel’s server CPU supply-demand mismatch as a surprise. It wasn’t. It was well-telegraphed across multiple earnings calls. Investors who thought Intel would capture more of the AI CPU demand surge simply weren’t paying attention to previous signals.

Yes, the call had a few silver linings that optimists will get excited about. But even then, understanding the timing of those events matters.

The Telegraphed Constraint

The supply shortage that “disappointed” investors in Q4 was explicitly stated on the two prior earnings calls.

Q2 2025 call (July 24, 2025), CFO David Zinsner:

“Capacity for Intel 7 remains very tight… We expect Intel Foundry revenue down slightly quarter-over-quarter due to capacity constraints in Intel 7, which we expect to persist through the second half of the year.”

Q3 2025 call (October 23, 2025), CFO David Zinsner:

“Capacity constraints, especially on Intel 10 and Intel 7 limited our ability to fully meet demand in Q3 for both data center and client products… shortage is pretty much across our business, I would say. We are definitely tight on Intel 10 and 7. Obviously, we’re not looking to build more capacity there. And so as we get more demand, we’re constrained. In some ways, we’re living off of inventory.”

Intel was clearly constrained on legacy nodes, living off inventory, and not building more capacity there. When Q4 AI CPU demand spiked, they couldn’t meet it.

Yes, like Stacy said, this is a shame. Yet it wasn’t a surprise. Here, let’s go a bit deeper into why Intel was tearing down 10/7.

Mix Shift

Intel wasn’t necessarily asleep at the wheel, but was executing a deliberate strategy to shift mix toward leading-edge nodes with better economics.

From the Q4 call, CFO Zinsner explained the logic:

“Panther Lake, while obviously it’s going to be a great cost structure for us over time given the wafers are fabbed internally—initially, obviously, when you got a new product on a new process, they’re pretty expensive products to start with. And so it’s dilutive in the beginnings of the year, and then it gets better over the course of the year... as we move towards more leading-edge mix, 18A for sure but also even Intel 4, 3, those products have better pricing and a better cost structure.”

Intel 4/3/18A have better ASPs and better cost structures than Intel 7/10. Intel was trying to push the mix toward these nodes. This is the correct long-term strategy if you weren’t expecting the AI CPU server demand spike.

The miss was Intel not forecasting the spike.

Tooting my own horn I bit… but I saw it coming.

Back in January 2025, I wrote in CPU Server Growth, Enterprise AI Agent Adoption Challenges:

“What’s the implication for the semiconductor industry? Access to more [AI] workers means more traditional work gets done, which increases the load on conventional and accelerated infrastructure.

This is a bull case for CPU-server TAM expansion!

An agentic world means more CPUs are needed because more workers are doing traditional CPU-based tasks.”

This was a bull case for CPU-server TAM expansion that the industry didn’t fully price in. A few industry execs reached out eventually, saying it was the first time they’d heard this argued—and that when they checked with their customers, it matched what customers were starting to see.

Stacy Rasgon from Bernstein hit the inventory issue on the head during the call:

“I mean, you guys have your own factories, like why are you in the inventory situation that you’re in?... I mean like you have $11.6 billion of inventory. And yet, it’s not in the right place at the right time to ship—like how does that happen?”

Zinsner’s answer was honest:

“If you go back six months or so ago and looked at what the outlook was, core count was absolutely looking like it would increase, but the units were not expected to increase. And every hyperscaler customer we talked to was signaling that. And obviously, it has rapidly increased over the third and fourth quarter.”

Intel, like the rest of the industry, forecast core count increases but not unit increases. When units spiked due to agentic AI demand, they had the wrong inventory in stock. That’s a miss.

Zinsner hints that they were basing forecasts on conversations with customers, so it sounds like an industry-wide miss.

But a miss for Intel nonetheless.

Advanced Packaging Timing Trap

The bullish statements on advanced packaging from the earnings call were honestly surprising and exciting. $1B+ deals and prepaying!

CFO David Zinsner:

“Another early indicator, I think, of success in the foundry is going to be advanced packaging. And we’ll start to see that revenue come in even before we start to see meaningful wafer revenue... I’d say some of the early customer engagements suggest that we’ll be well north of $1 billion on many of these opportunities for advanced packaging. So they’re way more exciting than even I had expected.”

CEO Lip-Bu Tan:

“I think the EMIB-T, I think, is a very big differentiator for us. And then clearly, we have a couple of customers willing to even prepay the substrate—because [substrate] is very big supply shortage and then they’re willing to share with us.”

Customers prepaying for advanced packaging capacity! EMIB-T is differentiated technology!

But investors not paying close attention will probably get ahead of themselves on this one too. It’ll be another timing issue.

Why? The shortage has moved upstream from advanced packaging (CoWoS capacity issues) to substrates.

Substrates have demand pressure too. Next-generation AI platforms require a massive substrate area. Thickness, or layer counts, are expanding too. AI server substrates are moving toward 20-30 layers to accommodate complex interconnects and maintain signal integrity.

A significant portion of the substrate shortage is directly tied to the availability of T-Glass (low-CTE glass cloth), a specialized material essential to maintaining dimensional stability in high-end designs. Nittobo currently dominates the T-Glass market with an estimated 80% market share. Demand for T-Glass is oversubscribed, and despite capacity expansion projects in Taiwan and Japan, supply is not expected to ease significantly until 1H27 or 2028.

Intel’s own 10-K points this out too:

current industry supply constraints for memory chips, substrates and foundry capacity that may impact our customers’ abilities to assemble products incorporating our products.

So, even though Intel’s advanced packaging deals are promising, upstream constraints may push out revenue timing.

Q1 2026: More of the Same

Intel has told us directly that 2H26-1H27 will be the exciting window with regards to foundry customers and presumably foundry capacity.

But don’t expect anything dramatic next quarter. CPU server supply will still be short. Yes, Intel is trying to meet demand, CFO David Zinsner on the call:

“We are aggressively getting tools on Intel 7, Intel 10, Intel 3, 18A, that is happening.”

But tools take time to install and ramp. So there likely won’t be a huge increase in CPU server supply in Q1.

I wouldn’t expect 14A customer announcements either.

And probably not a huge influx of advanced packaging capacity.

What to Watch

As we’ve been saying, indicators for Intel’s turnaround won’t come from earnings beats or customer rumor announcements. Watch for CapEx inflection from Ohio tool purchases and tool purchase signals at WFE suppliers, plus substrate capacity announcements in the supply chain. Obviously, keep watching Intel’s improving yields as a main KPI.

Yes, Intel’s long-term future still holds promise. It’s just going to take time.