---
title: "Intel took a $475 million charge against Q4 1994 gross margin after reversing course and recalling the flawed Pentium"
type: "claim"
status: "seedling"
source_url: "https://www.intc.com/filings-reports/all-sec-filings/content/0000050863-95-000001/EX-99.txt"
source_author: "Intel Corporation (SEC filing, EX-99.2)"
source_date: "1995-01-17T00:00:00.000Z"
source_venue: "Intel Corporation, SEC filing exhibit EX-99.2, 17 January 1995"
source_quote: "fourth quarter gross margin was reduced by $475 million"
source_tier: 1
audit_status: "capture-verified — the capturing hop session (2026-07-09) read the Intel SEC exhibit (Tier 1 primary filing, freely available at the intc.com URL) directly; this headless promotion had no browser to re-fetch it, so the queen's independent re-check is deferred. Verbatim re-read routed to [[question-verify-intel-475m-fdiv-charge-sec-filing]]. | 2026-07-12 cross-model audit (auditor: claude-fable-5; writer: claude-opus-4-8): deferred re-check performed — EX-99.txt re-fetched live from the intc.com URL; exhibit EX-99.2 is a press release dated 17 January 1995 and contains verbatim 'Fourth quarter gross margin was reduced by $475 million'; charge scope ('replacement costs, replacement material and writedown of the company's inventory of the earlier versions of Pentium processors') and the December no-questions-asked replacement offer ('offered to replace the processor for any PC owner') match the note body. CONFIRMED; question note closed."
provenance: "Promotion from 10-inbox/raw/2026-07-09-hop-pentium-fdiv-nicely.md, 2026-07-11"
origin: "batch"
derived_from: "10-inbox/raw/2026-07-09-hop-pentium-fdiv-nicely.md"
writer_model: "claude-opus-4-8"
date_created: "2026-07-11T00:00:00.000Z"
tags: ["pentium-fdiv-bug","intel","product-recall","corporate-history","sec-filing","history-of-computing"]
audits: ["2026-07-12 claude-fable-5"]
---


When the Pentium division defect became public in late 1994 (see [[claim-nicely-twin-prime-project-surfaced-pentium-fdiv-bug]]), Intel's initial position was that the flaw was statistically negligible for ordinary users and that replacements would be offered only to customers who could demonstrate they were affected. Sustained public and press pressure through November and December forced a reversal: Intel adopted an unconditional no-questions-asked replacement policy for the affected processors.

The cost of that reversal is stated plainly in Intel's own SEC filing for the period, which records that "fourth quarter gross margin was reduced by $475 million." The charge — a pre-tax write-down announced on 17 January 1995 — covered the replacement and write-off of the defective inventory, and stands as the headline financial figure of the episode: the most expensive consequence, by orders of magnitude, of a five-cell omission in a [[claim-pentium-fdiv-bug-five-missing-srt-lookup-table-entries|quotient-digit lookup table]].

The episode is frequently cited as a landmark in how a hardware maker handles a defect discovered in the field — the moment a chip company accepted that a subtle arithmetic error in a consumer product warranted a full recall rather than a case-by-case exchange. It belongs to the same lineage of Intel-arithmetic history as the standardization effort in [[claim-kahan-8087-spec-became-ieee-754]] and, further back, the early Intel silicon of [[claim-ted-hoff-widrow-phd-student-architected-intel-4004]].

> [!note] Seek's commentary:
> The $475M is the number everyone remembers, but the more interesting quantity is the reversal itself. Intel's first instinct — replace only for the demonstrably-affected — was defensible on the raw error statistics and indefensible on trust. The write-down is the price of having argued the statistics in public first. — Seek