Intel Is Growing Again, but the Stock Already Prices the Foundry Win

Intel just reported its fastest quarterly revenue growth since 2011, and the stock is down 38% from its high. Both camps sign the same record: seven straight quarters of beating guidance, a data-center business up 59% in a year, and a manufacturing unit whose yields have finally turned. The fight is over what that record is worth at $88.24 a share — after a 139% year-to-date rally, a $142.35 peak, and a $23 billion stock sale that new buyers are already under water on. One camp sees a durable AI cycle plus a foundry option that costs nothing more to hold. The other sees a price that demands two miracles, one of which has not yet billed a single meaningful dollar to an outside customer.

That second miracle is the whole argument. Intel's external foundry business generated roughly $293 million of revenue in the second quarter — about 5% of its $5.8 billion foundry segment, which lost ~$2.1 billion operating. At $88, the market is paying for that business to become real.

The Shared Record

As of August 26, 2026, per market data: price $88.24; market cap ~$466 billion, enterprise value ~$487 billion; 52-week range $23.68 to $142.35; shares outstanding roughly 5.3 billion after the August raise.

Q2 2026 (reported July 23):Revenue $16.13 billion, up 25% year over year from $12.9 billion, about 12% above consensus of ~$14.43 billion. Non-GAAP EPS $0.42 versus the ~$0.21 expected. GAAP EPS was a loss of $2.16 — an $11.0 billion net loss — dominated by a $12.5 billion non-cash mark-to-market charge tied to the U.S. Department of Commerce's escrowed CHIPS Act shares. Operating cash flow was $7.0 billion.

Segments: Data Center & AI revenue $6.3 billion, up 59%; Client Computing and Physical AI $8.9 billion, up 13%; Intel Foundry $5.8 billion, up 31%, of which ~$293 million was external (up from $174 million in Q1), with an operating loss of ~$2.1 billion (Q1 loss ~$2.4 billion).

Quality: Non-GAAP gross margin 41.8%, up 12.1 percentage points year over year; non-GAAP operating margin 17.2%. At the cash level: operating cash flow TTM ~$14.9 billion, capex TTM ~$12.1 billion, free cash flow TTM ~$2.8 billion; cash $12.9 billion, total debt ~$99 billion.

Guidance (management assertion): Q3 revenue $15.8–16.8 billion, non-GAAP EPS $0.38, non-GAAP gross margin 42% — above consensus. 2026 capex raised to above $20 billion; 2027 guided "significantly higher," with plans to tap capital markets. Management says demand exceeds its ability to supply and expects to remain supply-constrained into Q4.

The raise: On August 10, Intel announced a $15 billion offering, upsized the same day to $20 billion at $95, then sold the 15% overallotment, for roughly 242 million shares and about $23 billion. The stock now trades below the $95 price.

Foundry technology (company claims): 18A yields improving 7–8% per month; Panther Lake (Core Ultra Series 3) in high-volume manufacturing; 14A targeted for risk production 2028 and volume 2029; management expects external foundry customer commitments in the second half of 2026; multi-year agreements reported with Apple and Elon Musk's TeraFab, plus customer prepayments for substrates.

Round 1 — The comeback is real. The question is whether it compounds.

Bull's case. The product cycle is not a rumor; it is on the income statement. Revenue accelerated from $12.9 billion a year ago to $16.1 billion without a single quarter of help from the foundry's external business. Data Center & AI grew 59%, and management claims AI-related revenue is roughly 70% of the total, up over 70% year over year. Customers want more than Intel can produce — it says it has signed ten long-term server-CPU agreements with locked-in pricing or volume — and a global memory shortage that management expects to persist into 2028 is constraining supply, not demand. That is the rarest combination in a chip cycle: pricing power plus an order book that outruns your factories. Gross margin jumped 12.1 points to 41.8%, and operating margin went from negative to 17.2%. The machines are turning.

Bear's answer. The machines are turning, but off a crater. The 25% growth is measured against 2024–25, when Intel was losing share, bleeding cash, and cutting a dividend. Data Center & AI is still mostly a CPU story, and Intel's slice of the AI compute budget is a small fraction of Nvidia's $75.2 billion in data-center revenue in its latest reported quarter — so the CPU rebound is exactly the kind of cyclical snap that can peak within a year or two rather than compound. Management itself says client revenue will be roughly flat in Q3 because memory shortages cap shipments — so the growth engine is partially throttled by a shortage it does not control, and the "won't catch up in Q4" line means revenue will keep being pushed into future quarters, flattering near-term bookings. A 25% growth quarter is evidence of a cycle; the evidence that it lasts three years is still just guidance.

Winner: The bull wins this round on what has already happened, and the bear wins the durability question. For beginners this is critical: Intel's turn is measured and real, but a single great quarter does not establish that growth compounds.

Round 2 — Foundry is fixed on the factory floor, not on the income statement.

Bull's case. For three decades the bear case on Intel's foundry was yields. That excuse is gone. Management reports 18A yields improving 7–8% per month, Panther Lake is in high-volume manufacturing on 18A with ASML High NA EUV, and the 0.5 process design kit is out with a 0.9 on the way. 14A is timed near TSMC's 1.4nm node — risk production 2028, volume 2029. Most telling, customers are prepaying for substrates, and multi-year agreements are reported with Apple and TeraFab; expects external commitments to land in the second half of this year. The CEO calls the foundry a "national treasure," the U.S. government owns roughly 10% through the CHIPS Act escrow, and wafer demand exists that Intel could not satisfy even for itself a year ago. The option is real, and at this point it is funded.

Bear's answer. The option has a strike price and no revenue. In Q2 the entire external foundry business was ~$293 million — about 5% of the segment — against a ~$2.1 billion operating loss. Intel has run this play before: it has tried for years to be a merchant foundry and never converted customer interest into meaningful outside revenue. The named wins so far (Apple, TeraFab) are reports and agreements, not revenue on the income statement, and no leading-edge external design win has been disclosed in the results. Meanwhile, the foundry is where the money goes: 2026 capex is now guided above $20 billion, 2027 significantly higher, funded in part by a $23 billion stock sale. The bear does not need yields to fail; it only needs commitments to slip, and each quarter of ~$2 billion in foundry operating losses is roughly $0.40 a share that the stock's ~$1.45 annualized EPS run rate must absorb. When your option is priced as if exercised, it is no longer an option.

Winner: The bear wins the economics round on current facts. The bull's counter is a date — H2 2026 commitments — which is exactly the sort of promise a verdict should need to survive.

Round 3 — What $88 demands, and who pays for it.

Now make both stories pay rent in the same units. Intel's non-GAAP EPS is tracking an annualized run rate of roughly $1.45 (Q1 actual $0.29, Q2 actual $0.42, Q3 guide $0.38). At $88.24, that is about 60 times annualized earnings. On any reasonable valuation frame, the market is not paying for this year; it is paying for roughly a doubling of EPS to ~$2.94 just to reach a still-generous 30 times, or a tripling to ~$4.40 to reach a mature 20 times.

The bull reads that as cheap versus Nvidia at 16.7x sales and AMD at 19x; Intel stands at 8.2x revenue and 39.6x EV/EBITDA. The bear notes Intel is the only one of those names that loses money on a GAAP basis — and its low sales multiple is the arithmetic of a ~42% gross margin against competitors' far higher ones, not a free lunch.

Then there is dilution. Intel sold ~242 million shares — about 5% of the company — at $95 to fund the foundry and capacity buildout, and it said it will keep tapping capital markets. The stock now trades near $88, below that price, meaning last week's buyers are already losing money and the company raised its most expensive (equity) capital at the worst of the recent price action. The 2027 capex plan makes more issuance likely. Bulls are right that debt is not rising and the rating is protected; bears are right that when cash flow cannot cover the buildout, per-share earnings are bigger on paper than in your pocket.

One more counterintuitive line item belongs to the bear. Because the U.S. government's CHIPS shares are held in escrow and rise in value with the stock, Intel must mark that liability up in its GAAP results — the $12.5 billion charge that turned a strong operating quarter into a reported $11 billion loss. The company does not lose cash, but the government's ~10% stake gets more expensive to shareholders on every rally. A rising stock mechanically produces a reported loss. That is a headwind most casual readers of the headline earnings number will never see.

Winner: The bear wins the price round. At $88 the market has already financed the foundry, diluting holders by ~5% in the process, and still needs EPS to double before the multiple looks normal.

The Verdict

The business case goes to the bull. The operating record — seven straight beats, 25% revenue growth, margins re-expanding, demand exceeding supply — is the strongest evidence Intel has produced since the turnaround began, and it belongs to both the product line and a genuinely improved manufacturing operation.

The stock call goes to the bear at this price. $88 is above fair value on the earnings visible today, and it only becomes fair or cheap if the foundry's external business converts from agreement to revenue on a timeline the price is already using. That reverses the burden of proof: it is not the bears who must show the turnaround fails, but the bulls who must show it compounds at roughly twice today's rate while absorbing the ~5% dilution already sold and whatever further equity the 2027 buildout demands. A business can be healthy and its stock fully priced — this is that place.

The reversal clause. The ruling flips to bullish if, by the Q1 2027 report (~April 2027), two things are visible: a named external foundry customer on the leading-edge (14A) with committed volume, and external foundry revenue scaling materially past the ~$500 million-a-quarter range, with foundry operating losses trending toward $1 billion or below. That would be evidence the option is converting, and 60-times-annualized earnings would look less expensive.

The ruling is confirmed bearish if foundry losses stay near $2 billion a quarter through the end of 2027, if the H2 2026 external commitments fail to convert to disclosed revenue, or if Intel needs another large equity raise before its own data center growth slows even slightly.

Watch next:Intel reports Q3 on ~October 22, 2026. The bull's earliest confirming signal is the same as the quarter's headline: management says it remains supply-constrained and gross margin holds near 42%. If shipments get throttled and foundry losses widen instead, the market has its answer about whether $88 was the high or the halfway point.