The Intel turnaround — what changed and what hasn’t

Intel in 2026 is a company fighting for its survival as a tier-one semiconductor company. That statement would have been unthinkable in 2015, when Intel commanded 90%+ of the server CPU market, was considered unassailable in PC processors, and had a manufacturing technology lead over every competitor. What followed was one of the most dramatic competitive collapses in technology industry history — driven by a combination of manufacturing execution failures, strategic miscalculations, and the emergence of NVIDIA’s GPU as the primary AI computing platform Intel never anticipated.

The key inflection point was the 10nm delay (2016–2019) — when Intel couldn’t deliver its next-generation manufacturing process on time, leaving AMD’s TSMC-fabricated Ryzen chips to compete at parity or better in performance per watt. That opened the door for AMD to take meaningful market share in desktop PCs, laptops, and then — crucially — data center servers, where AMD’s EPYC chips have gained 30%+ market share from near-zero in 2018.

The 18A process node — Intel’s everything bet

Intel’s 18A process node (the “A” stands for Angstrom — 1.8 angstroms, a reference to the gate-all-around transistor dimensions) is the most important product in Intel’s history since the original microprocessor. If 18A delivers competitive yields at high volume, Intel’s foundry business becomes credible and the entire turnaround thesis is validated. If 18A fails, Intel Foundry may need to be abandoned or spun off, and Intel becomes a pure fabless chip designer dependent on TSMC — a fundamentally smaller and lower-margin business.

The technical details: 18A uses RibbonFET (Intel’s name for gate-all-around transistors) and PowerVia (backside power delivery) — two genuinely innovative manufacturing techniques that, if implemented successfully, could provide transistor density and power efficiency comparable to TSMC’s N2 (2nm) process. Independent assessments from chipmakers who have received 18A test wafers have been cautiously optimistic: yields are improving but haven’t yet reached the threshold needed for high-volume external customer commitments.

Panther Lake — the first 18A product

Panther Lake is Intel’s first processor built on the 18A node, targeting high-performance laptops and the AI PC market. A successful Panther Lake launch in 2025–2026 serves as the proof point for 18A manufacturing capability — the same node Intel Foundry would use to manufacture chips for external customers. Early performance data has been encouraging: Panther Lake reportedly shows significant IPC (instructions per clock) improvement over current-generation Intel CPUs and best-in-class power efficiency for the laptop segment.

Intel Foundry — the $100B+ gamble

Intel Foundry Services (IFS) — Intel’s external chipmaking business — remains the largest single bet in Intel’s turnaround strategy. The premise: the United States government, alarmed by 90%+ semiconductor manufacturing concentration in Taiwan (geopolitical risk) and South Korea, has committed $52 billion via the CHIPS Act to rebuild domestic semiconductor manufacturing. Intel, as the only U.S.-based company capable of building leading-edge semiconductor fabs, is the primary beneficiary.

Intel has been awarded approximately $8.5 billion in CHIPS Act grants plus $11 billion in CHIPS Act loans — making the U.S. government essentially a partner in Intel’s manufacturing resurrection. The geopolitical tailwind is real and durable: as long as Taiwan remains a geopolitical flashpoint, U.S. policymakers have an existential incentive to ensure Intel’s manufacturing succeeds.

The challenge: Intel Foundry has been generating billions in quarterly losses while building out fab capacity that doesn’t yet have sufficient external customers to fill. In 2026, the critical business development question is whether Amazon Web Services, Microsoft Azure, or another hyperscaler commits to using Intel 18A for their custom chip programs. Such an announcement would be the single most bullish catalyst possible for INTC.

PC and Server CPU — the core business under pressure

Intel’s bread-and-butter business — x86 CPUs for PCs and data center servers — remains its largest revenue source but is under structural pressure. In the PC market, AMD’s Ryzen chips have taken meaningful share in the high-performance enthusiast and creator segments. In laptops, Apple’s M-series chips (made by TSMC) have demonstrated that ARM-architecture processors can dramatically outperform Intel on performance per watt — and Windows on ARM is gaining mainstream traction.

In the server market, AMD’s EPYC “Turin” chips (on TSMC 3nm) are widely considered performance equals or superiors to Intel’s Xeon. Cloud providers — Amazon, Microsoft, Google — are increasingly designing their own custom ARM-based processors, reducing dependence on x86 entirely. Intel’s Xeon “Granite Rapids” and upcoming “Diamond Rapids” are competitive products, but the tailwind of being the only viable option is gone.

The Lip-Bu Tan era — what’s different

Lip-Bu Tan has made clear that he views Intel’s challenge as a focus and execution problem, not a strategy problem. His early moves: cutting 15,000+ jobs, canceling or delaying non-core projects, rebuilding engineering leadership, and instituting a more disciplined capital allocation framework. He has also signaled openness to strategic partnerships — including the possibility of allowing TSMC to manage Intel’s existing fabs or partnering with external foundry customers on 18A development. The contrast with Gelsinger’s era (which was characterized by bold proclamations) is a more engineering-focused, results-first culture.