Intel reported third-quarter 2026 revenue of $16.5 billion, up 21% year over year, with growth in data center chips offsetting a sharp contraction in PC sales and a fab outage that cut into supply. For teams planning on-prem or edge inference builds around Intel Xeon parts, the quarter is a preview of tighter availability and higher system costs into early 2027.
The Data Center and AI group generated $7.3 billion in revenue, up 78% year over year, on demand from hyperscale, enterprise and cloud customers. But Intel said it could not meet that demand for its Xeon server family, citing shortages of raw materials, substrates and other critical components. A two-week shutdown at an Intel fabrication facility in Israel, triggered by a gas explosion, compounded the shortfall.
The fab math matters for anyone sequencing a hardware refresh: EE Times reports that turning a silicon wafer into a finished chip takes roughly three months, so wafers still in process when the Israel fab shut will likely be scrapped rather than recovered. Intel CFO David Zinsner said on the earnings call that most of those write-offs will land in the fourth quarter, with no further scrapping expected in the first quarter of 2027. He put the fab disruption's revenue impact at an estimated $300 million in Q4 2026, with a similar hit in Q1 2027.
Intel's near-term guidance reflects the squeeze: fourth-quarter revenue is projected at $15.0 billion to $16.0 billion, with non-GAAP gross margin forecast at 38.0%, down from 44.1% in the third quarter. CEO Lip-Bu Tan told the earnings call, "I'm disappointed that we were not able to meet strong customer demand in this quarter. We face near-term challenges, including manufacturing supply shortage, the unprecedented memory crunch, and the resulting softness in PC demand."
The memory side of the equation hits architects planning edge or on-prem inference just as hard as CPU scarcity. Rising DRAM prices from Samsung, SK Hynix and Micron, driven by AI infrastructure demand, pushed PC makers to raise retail prices, and Intel's Client Computing and Physical AI Group saw revenue decline 5% year over year to $8.1 billion. Samsung and SK Hynix both reported record quarterly profits over the same period, underscoring that the component crunch is a net win for memory vendors even as it raises build costs downstream.
Counterpoint Research data cited by EE Times shows global PC shipments fell 21% year over year in the third quarter to 57 million units, with the firm forecasting a further 27% year-over-year decline in the fourth quarter. Counterpoint senior analyst Minsoo Kang said, "The global PC market is facing mounting pressure as rising memory and component costs push system prices higher and force consumers to delay replacement purchases." Counterpoint associate director David Naranjo added that "rising component costs, higher system prices, and longer replacement cycles are fundamentally reshaping market dynamics."
What's unresolved is how long this persists. Tan told the earnings call, "We expect the current environment to persist until memory availability and prices begin to normalize," and said "the next few quarters will be challenging for the PC market." Zinsner said Intel does not expect positive free cash flow in 2027, calling it an "investment year" during which the company plans to raise capital expenditure to rebuild production capacity. None of that is a fixed timeline — it's a company saying its own supply will stay constrained while it spends to fix it, with a specific $300 million revenue hit already booked for Q4 2026 and Q1 2027 each.
For architects sourcing CPUs or planning memory-dense inference nodes, the takeaway is to lock in server-grade DRAM and Xeon allocation now, with contracts that price in scarcity rather than assuming current list prices, because Intel itself is only committing to a quarter-by-quarter view of when memory and component costs normalize.