The one number
5.1%.
That is the share of Intel Foundry's Q2 segment revenue that came from external customers: $293 million out of $5.765 billion. Most Foundry activity still served other Intel businesses. The filing also says the increase in external revenue was primarily driven by Altera becoming an external customer following its deconsolidation. Q2 Form 10-Q, segment note and Foundry discussion.
That makes today's Terafab headline relevant. Investors are watching whether Intel's manufacturing technology can win substantial outside business. They need to distinguish an opportunity, an agreement and revenue earned from it.
We last reviewed Intel on September 20. This return to the company examines the new headline against the operating evidence and the recovery since August. Earlier Intel review.
Why Intel is back in the headlines
The October 5 Investing.com report carried by Yahoo says TSMC is discussing potential involvement in Elon Musk's Terafab venture. It attributes confirmation of early discussions to Musk. Intel's April earnings materials had already identified Intel as a strategic partner in the project. Today's report; Intel's April remarks.
The possible involvement of another manufacturer changes the range of outcomes investors must consider. It does not establish that Intel has lost a signed order, or that its role has been cancelled. No finalized TSMC agreement is established by this report. My read: the immediate issue is uncertainty over Intel's eventual role and economics.
The report also relays Susquehanna's concerns about Intel's PC processor model share. A count of available processor models is not a count of chips sold or a measure of recognized revenue. That research is another item to investigate against Intel's own shipments, sales and product-ramp disclosures, rather than a substitute for them.
The rebound is real; the averages still carry the earlier decline
Intel closed Friday, October 2 at $119.33. We reproduced the daily simple moving averages from closing prices: 50-day $100.87; 100-day $108.35; 200-day $81.18. Price was above all three. The supplied chart draws the 100-day curve and a horizontal line at the latest 50-day value; its amber curve is not the 200-day. Daily prices used for verification.
Friday's close was 40.2% above the August 24 intraday low of $85.14, but 16.2% below the June 30 intraday high of $142.35. These describe different parts of the same path. The 50-day remained 6.9% below the 100-day, a relationship that began September 1.
Moving averages lag: the rebound can lift price above them before the shorter average catches the longer one. An eventual crossover would describe improving historical prices, not guarantee future returns. Requiring that crossover as the sole proof of recovery would miss the price improvement already visible.
The attached 7:51am ET October 5 premarket snapshot is $114.45, down 4.1% from Friday's close and still above the three Friday averages. It is an extended-hours observation, not Monday's close or a live quote. Friday itself opened at $124.01 and finished near its session low; the close was about 5.3% of the way up its high-low range, rounded to 5% on the supplied card.
The question now is whether price holds stronger lows and recovered ranges as the new information is absorbed. The nearer references are the Friday 100-day around $108 and 50-day around $101. They move as new sessions enter the calculation; they are monitoring references, not prescribed entries or automatic support.
The foundry headline and the foundry accounts
Making chips for Intel's own products remains economically important. It builds manufacturing experience and serves an existing product business. But internal activity does not demonstrate the same customer franchise as a broad outside foundry operation.
Intersegment sales are eliminated in the consolidated accounts. Adding the Foundry segment's revenue to Intel's product revenue would overstate Intel's total sales.
Altera matters because a transaction can change the category in which revenue appears without creating an entirely new customer relationship. External revenue growth therefore needs a customer and product explanation, not just a percentage increase. The next useful evidence would be committed demand, production milestones and recognized revenue with the economics attached.
What has improved in the reported business
Q2 consolidated revenue increased from $12.859 billion to $16.128 billion, up 25.4%. Foundry's operating loss narrowed from $3.168 billion to $2.089 billion. The quarter shows improvement, with Foundry still loss-making. These measures have different business scopes; a smaller segment loss is not the same as a profitable external foundry. Q2 release, income statement and segment results.
The filing attributes most of the Foundry loss reduction to $1.4 billion of lower period charges, including the absence of earlier impairment and accelerated-depreciation charges. That means the entire improvement should not be presented as recurring factory efficiency. The tension is worth following: stronger product demand and fewer charges can improve today's accounts before the economics of a broad external foundry business are proven. Foundry loss explanation.
Why the GAAP loss and adjusted earnings tell different stories
Intel reported $1.796 billion of GAAP operating income, but a $11.033 billion net loss attributable to Intel and GAAP diluted EPS of minus $2.16. Company-defined non-GAAP diluted EPS was $0.42. The earnings definitions differ. Q2 release and reconciliation.
A $12.5 billion escrow-share fair-value loss, driven by Intel's rising share price, affected the GAAP result. The cash-flow statement reverses that accounting charge in reconciling net income to operating cash flow. That helps explain why the net loss alone is an incomplete account of operating performance. It does not remove spending, dilution or funding obligations. Q2 filing, escrowed shares and cash flows.
The valuation needs a history, not two snapshots
At Friday's close, Intel traded at 109.5 times the sum of its last four reported non-GAAP quarterly diluted EPS figures: $0.23 + $0.15 + $0.29 + $0.42 = $1.09. The equivalent GAAP sum was minus $2.11, so a positive trailing GAAP P/E is not meaningful. This is a trailing adjusted measure, not a forward earnings forecast. Intel's quarterly releases.
The graphic follows 523 daily closes from September 2024 through October 2, 2026. Each earnings denominator changes only in the session after Intel released the new quarter. That avoids putting information into a historical valuation before investors could have known it. Where the adjusted earnings sum was negative—January 31 through October 23, 2025—the chart shows N/M, not a misleading positive multiple.
Watch the July reset. On July 23, the $100.23 close divided by $0.57 of trailing adjusted EPS produced 175.8×. On July 24, after the new earnings release, price fell to $92.32 but the earnings sum rose to $1.09: 84.7×. The fall in P/E reflected both a price decline and a larger denominator. Since then, the rebound has lifted that multiple to 109.5×. A lower multiple cannot, by itself, tell us that a stock has bottomed or become inexpensive.
The earlier negative-earnings interval also matters: rolling the impaired Q3 2024 quarter out of the four-quarter window helped the sum turn positive. Not every improvement in trailing EPS represents newly earned operating profit. Non-GAAP adjustments, tax assumptions and the Altera business scope changed across this period.
These are sums of rounded quarterly EPS, rather than a reconstructed annual EPS figure; share weighting can make the two differ. The history is useful for understanding the moving denominator, but it does not establish Intel's cheapest valuation ever, a fair price, or a valuation ranking against companies using different earnings definitions.
The next checkpoint
Watch what Intel discloses about Terafab's scope, customer commitments and production economics. Treat third-party accounts of discussions as reports until the companies establish what has actually been agreed.
Then revisit outside Foundry revenue and its customer mix, manufacturing margins and the next reported cash position. Intel's July Q3 revenue guidance was $15.8 billion–$16.8 billion; this is a dated company forecast, not a completed quarter's result. July guidance.
For the stock, look at the durability of the rebound alongside those results. A negative morning headline does not erase six weeks of price improvement; a 40% rebound does not prove every expectation embedded in the valuation will be delivered.
The question: can Intel turn a recovering product business and manufacturing progress into outside customer economics that justify the stock's expectations?
Educational commentary, not a buy or sell recommendation. Publication: October 5, 2026. Regular-session market basis: October 2 close. Premarket observation: October 5, 7:51am ET. Reported financials through June 27, 2026; later developments are separately dated.