A fatal explosion at Intel's Fab 28 plant in Israel has knocked out two weeks of 10nm-class chip production and will squeeze Xeon supply well into 2027. For UK buyers already fighting a brutal memory shortage, this is another reason to track server lead times and supply chains closely.
View the data behind this chart
| Revenue impact ($m) | Q3 2026 | Q4 2026 | Q1 2027 |
|---|---|---|---|
| Revenue impact | 0 | 300 | 300 |
What happened at Fab 28
Intel confirmed that an explosion at its Fab 28 facility near Kiryat Gat, Israel, earlier this month killed two people and seriously injured another, forcing a full two-week shutdown of the site. Intel's chief financial officer, Dave Zinsner, told investors on the company's earnings call: "In Israel, we have been shut down for about two weeks... we feel like we are going to be starting up production here by the end of this week."
Fab 28 is one of Intel's core manufacturing sites for chips built on its 10nm-class process technologies, now branded Intel 7 (formerly known as 10nm Enhanced SuperFin). That makes it a critical node for server CPU output, not a peripheral facility that can simply be worked around overnight.
Which Xeon products are exposed
Intel has not named the specific Xeon families affected, nor disclosed how many wafers were lost, but the exposure is structural rather than incidental. Xeon 6 processors use newer Intel 3 compute dies, but their I/O dies are still fabricated on Intel 7 — meaning even Intel's current flagship server line depends on Fab 28-class capacity.
Older generations are even more exposed. Xeon 5 'Emerald Rapids' is built entirely on Intel 7, and Xeon 4 'Sapphire Rapids' — which Intel stopped accepting new orders for in late September — also relies on the same node for any remaining committed shipments. UK buyers running mixed fleets with Emerald Rapids or legacy Sapphire Rapids deployments should treat continuity of supply as genuinely uncertain, not just slower than usual.
Why Intel can't simply reroute production
Intel does operate other fabs capable of 10nm-class output, including Fab 12, Fab 22, Fab 32 and Fab 42 in Arizona alongside Fab 28 itself. In theory that sounds like redundancy. In practice, Intel has said it has already done everything feasible to rebalance output, and it still can't fully offset the Fab 28 gap.
Demand for server processors is already running hot, qualification of specific products at alternative fabs takes time and money, and Intel will not repurpose the lines making Xeon 6 I/O dies to instead produce older-generation chips. The upshot: this is a genuine capacity shortfall, not a logistics inconvenience Intel can quietly absorb.
The financial scale, and why Q4 matters most
Intel expects the incident to cost roughly $600 million in lost revenue across two quarters — around $300 million in Q4 2026 and a similarly sized hit in Q1 2027. It also expects a roughly 2% reduction to non-GAAP gross margin in Q4 2026 specifically, concentrated there because most wafer write-offs from work-in-progress stock caught mid-fabrication during the explosion will land in that quarter.
Zinsner was explicit that "almost all" of the Q4 impact is tied to 10nm server products, which should reassure buyers of Intel's newer Intel 3-based parts somewhat — but offers little comfort to anyone depending on Intel 7-based Xeon 6 I/O dies or older Xeon generations.
What UK buyers should do now
With Fab 28 only just restarting and full capacity recovery undated, UK procurement teams should treat the next two quarters as a live supply-risk window rather than wait for clarity that may not arrive quickly. Locking in confirmed orders now, before Q1 2027 allocation tightens further, is the single most practical step.
It's also a sensible moment to stress-test the roadmap. Teams that haven't already should compare Xeon with EPYC alternatives as a hedge, and more broadly assess alternative server processors across vendors rather than assuming Intel capacity will simply catch up. For fleets that can tolerate a generation step-down, it's worth asking whether to decide whether to buy servers now on existing stock rather than waiting on new-build Xeon 6 or Emerald Rapids allocations.
Buyers with budget flexibility should also look at server finance options to secure pricing ahead of potential allocation-driven increases, and consider whether to configure a Dell server or configure an HPE server now while current-generation stock is still moving, rather than resubmitting specs later into a tighter market. Organisations less tied to brand-new silicon could also explore refurbished servers as a bridge while 10nm output recovers.
View the data behind this chart
| Phase | Starts (week) | Duration (weeks) |
|---|---|---|
| Two-week shutdown | 0 | 2 |
| Restart & ramp-up | 2 | 6 |
| Extended Xeon shortage | 8 | 18 |
What to watch next
Three things will determine how this plays out for UK buyers. First, whether Fab 28 actually returns to full output on schedule, or whether ramp-up slips further. Second, whether Intel discloses which specific Xeon SKUs and wafer volumes were affected, which would let buyers map risk to their exact fleet rather than the whole 10nm category. Third, whether the disruption spreads beyond servers — Intel has not ruled out impact on client CPUs built on the same Intel 7 node, including 12th, 13th and 14th Generation Core processors, though it hasn't confirmed Fab 28 makes any of these in volume.