Intel CEO Lip-Bu Tan said 18A yield is now 70%, up from “horrible” when he took over, in an interview with Gavin Baker at the Tomorrow X Summit on 18 November 2026, according to posts by attendees. The same day, Intel and Microsoft announced that Microsoft is the first company to announce use of Intel Foundry’s EMIB advanced packaging for Azure chips.
Our number: at 70% yield, about 30 of every 100 chips on an 18A wafer still do not make the cut (our arithmetic; how Intel measures yield was not described in the posts we read).
Business Pill · Vertical Integration
An evergreen Business Engineer explainer on vertical integration. It teaches the general idea only and does not discuss Intel or Microsoft.
The key insight: Lip-Bu Tan put Intel’s 18A yield at 70% and, per accounts of his remarks, said Intel signed its first outside 18A customer; Microsoft became the first company to announce use of Intel’s EMIB packaging, for Azure chips, with multi-year Xeon supply.
What Tan Said on Stage
Tan’s remarks reached us through attendees and market accounts on X, not a transcript. Fireside Alpha quoted him: “When I took over, 18A yield was horrible. Now we are at 70%”. An attendee, Anni Sen, wrote that “internal yields on 18A are at 70%” and called it “the number that matters”.
On the next node, Fireside Alpha quoted Tan: “Our first 14A product comes in 2027. Outside customers come in 2028, 2029”, and added: “I think realistically that’s about the same time as TSMC, maybe slightly behind”. Other accounts tie the TSMC comparison to Intel’s own chips: Wall St Engine reported Tan seeing parity with TSMC for Intel’s own chips by late 2027-28, and Sen’s notes put 14A’s internal use in late 2027 into 2028.
Wall St Engine also reported Tan saying “we just signed our first outside customer on 18A”, and an account quoted by P Equity Research wrote that “he announced that we have signed the first external customer on 18A”. Intel had not named the customer or issued a statement on it in the material we read.

Update, 18 November 2026, 20:43 UTC: Tan on AI demand. In a 20:18 UTC post on the same Tomorrow X Summit talk, Wall St Engine quoted Tan: “I just cannot supply enough” (post). Per that account, Tan expects more AI capex in 2028 than in 2027, though spending may shift as financing gets harder, and said the build-out could slow a bit after the midterms and that some correction is healthy, but overall he sees no slowdown. On agents he said: “Agents are a little bit like teenagers. They just go and do things”. As with the remarks above, we have seen no transcript or video; this is Wall St Engine’s account. The financing caveat echoes the Fed minutes released today, which flagged fast-growing AI debt (our report).
What Microsoft Signed
The joint release says the companies reached a long-term agreement covering “select areas of Azure infrastructure”. The companies say Microsoft is the first to announce use of EMIB, Intel’s Embedded Multi-die Interconnect Bridge, for “Azure first-party silicon”, and the agreement confirms multi-year Intel Xeon supply capacity, including next-generation server CPUs codenamed Diamond Rapids.
The release also cites up to 2.3x more performance on Azure M-series v4 virtual machines using CXL memory with Xeon 6, and up to 2.5x faster SQL backups with Intel QuickAssist. It does not give a contract value, a volume or a start date, and it does not mention 18A.
Connecting the Coverage
This sits on a run of Intel moves we have tracked: Intel retaking 70% of desktop chips, Tan’s $19.7 million of personal stock purchases and SoftBank’s Intel stake. Our packaging bottleneck explainer sets out why EMIB and CoWoS capacity matter, and Microsoft’s own chip push showed up in its Cobalt 200 pricing.
The Structural Read
As we read the release, Microsoft is buying Intel’s packaging and CPUs, not its leading-edge wafers: the agreement covers EMIB and Xeon supply and does not mention 18A.
As we read the accounts of Tan’s remarks, the 14A timetable he gave, first product in 2027 and outside customers in 2028-2029, puts Intel’s own products roughly a year or more ahead of foundry customers on the node.
Lip-Bu Tan, Tomorrow X Summit, 18 Nov 2026, as quoted by Fireside Alpha
“When I took over, 18A yield was horrible. Now we are at 70%”
Three Implications
FOR INTEL FOUNDRY The joint release makes Microsoft the first company to announce use of EMIB for its own Azure silicon.
FOR TSMC Per Wall St Engine, Tan sees parity with TSMC for Intel’s own chips by late 2027-28.
FOR AZURE The release confirms multi-year Xeon supply, including Diamond Rapids, and CXL memory work on M-series v4 VMs.
The Business Engineer Lens
This story maps onto the Business Engineer framework The AI Supercycle.
The analysis puts it this way: “Allocation power at the foundry is now more consequential than chip design itself.”
As we read today’s news, Intel is trying to sell that allocation power, in packaging first. The analysis predates these announcements and does not discuss them.
What Is Not Established
We have not seen a transcript or video of the interview, so Tan’s words come from posts by people in the room and market accounts. The 18A customer is unnamed, and we found no Intel statement confirming it. A 70% figure does not say which products, die sizes or test criteria it covers, and we found no published Intel 18A yield data. The Microsoft release states no contract value.
The Bottom Line
In one day Intel’s foundry got a named packaging customer in Microsoft and, per accounts of Tan’s remarks, an unnamed first outside 18A customer and a 70% yield figure, with 14A products in 2027 and outside 14A customers in 2028-2029. This is not investment advice.
97,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
A note on sourcing. Deal terms and performance figures are from the Intel and Microsoft joint release of 18 November 2026, read in full. Tan’s remarks come from X posts by attendees and market accounts at the Tomorrow X Summit; we did not see a transcript or video. The 30-in-100 figure is our arithmetic from the 70% yield. We did not contact Intel or Microsoft.
Sources: Wall St Engine on X: Tan on AI demand (18 Nov 2026) · Intel and Microsoft, joint press release, 18 Nov 2026 · Fireside Alpha on X, 18 Nov 2026 · Anni Sen on X, attendee, 18 Nov 2026 · Wall St Engine on X, 18 Nov 2026









