Intel INTC stock prediction: $148 bull case vs $62 bear case
Intel did not lose $11 billion last quarter. That number is real, it is audited, and it sits at the top of the company’s own Form 10-Q for the quarter ended 27 June 2026 — and it is still one of the most misleading figures in large-cap technology. The loss was caused almost entirely by a $12.5 billion mark-to-market charge on shares Intel owes the US government, a liability that grows every time Intel stock goes up. Intel shares closed at $97.71 on 11 August 2026, up 348% from their 52-week low of $20.44 and down 31.4% from a 52-week high of $142.35. Our 12-month range is a $148 bull case and a $62 bear case, and the single most useful thing an investor can understand about Intel right now is that its reported earnings and its share price now move in opposite directions by construction.
That is not a rhetorical flourish. It is arithmetic, and it is the reason so much Intel commentary is unusable. Under the agreement Intel signed with the Department of Commerce on 22 August 2025, 158.74 million shares sit in escrow and are released to the government as Intel hits Secure Enclave milestones. Intel must carry that obligation as a derivative liability and remeasure it at fair value every quarter. As of 27 June 2026, 143 million escrowed shares had not yet been released and the liability had swelled to $15.6 billion, up from $2.7 billion at the end of 2025. Do the multiplication yourself: at 143 million shares, every $10 move up in Intel’s share price adds roughly $1.4 billion of GAAP loss to the income statement. If our bull case lands and INTC reaches $148, Intel books about $7.2 billion of additional accounting loss on the way there. If the bear case lands at $62, Intel books a roughly $5.1 billion accounting gain. Intel is, in the most literal balance-sheet sense, short its own stock.
Key facts
- Spot price $97.71, close of 11 August 2026; 52-week range $20.44–$142.35 — stockanalysis.com
- Q2 FY2026 revenue $16.13 billion, up 25% year over year from $12.86 billion — Intel Form 10-Q, 27 June 2026
- GAAP net loss $11.03 billion (−$2.16 per share), driven by $12.5 billion of fair-value losses on escrowed shares — Intel Form 10-Q
- Intel Foundry external customer revenue $293 million in Q2, against a segment operating loss of $2.09 billion — Intel Form 10-Q
- Equity raise priced at 210,526,315 shares at $95.00, net proceeds ~$19.7 billion, closing 12 August 2026 — Intel investor relations, 11 August 2026
- Cash and short-term investments $29.73 billion against total debt of $50.54 billion pre-raise — Intel Form 10-Q
- 2026 capex guided to more than $20 billion, with 2027 “significantly above” that — CFO David Zinsner, Q2 earnings call, 23 July 2026
What actually happened in Q2 — and why the stock fell 7.9% on a blowout
Strip out the escrow accounting and Intel’s second quarter was the best operating quarter the company has printed in more than a decade. Revenue of $16.13 billion was up 25% year over year, the fastest growth rate Intel has recorded in over 15 years, against a consensus that sat around $14.4 billion. GAAP operating income swung to a positive $1.80 billion from a $3.18 billion loss a year earlier. GAAP gross margin came in at 40.4%, up 12.9 percentage points. Adjusted earnings of $0.42 per share came in at roughly double the $0.21–$0.22 the street had modelled.
The segment detail is where the story sharpens. Data Center and AI revenue rose 59% to $6.26 billion and — the number nobody quotes — its operating margin went from 16% to 40%, delivering $2.47 billion of operating income against $633 million a year earlier. Client Computing and Physical AI, renamed this year from Client Computing Group, did $8.88 billion at a 26% margin. Together, Intel Products earned $4.82 billion of operating income in a single quarter.
So why did the shares close at $92.32 on 24 July, down 7.9% from $100.23 the day of the print? Because the bar was never the beat. Intel entered that report already 28.9% below its 22 June closing peak of $140.94 and still up 171% year on year, and the thing the marginal buyer was waiting for — a named, marquee external customer for Intel Foundry — did not arrive. A 25% revenue beat does not clear a bar that was set somewhere else entirely. Our own coverage of the print, Intel beat Q2 earnings estimates twice over and its own rally cost it $12.5 billion, made the same point on the day.
Management was, to be fair, precise about the manufacturing progress. “Our factories across Intel 7, Intel 3, and Intel 18A exceeded internal volume targets,” CEO Lip-Bu Tan told analysts, adding that “18A output increased meaningfully in the quarter. Yields continue to track ahead of expectations,” per the Q2 2026 earnings call transcript.
The pricing-power problem hiding inside a 59% growth number
Here is the data synthesis that changes how the bull case should be underwritten. Intel’s own 10-Q breaks the Data Center and AI increase into its components, and the split is not what a 59% headline implies. Server revenue rose $2.0 billion in the quarter on ASP increases of 48%. Server volume rose 9%.
In other words, roughly five-sixths of Intel’s celebrated data-centre acceleration is price, not units. And Intel is explicit about why it could take that price: “Market demand exceeded our available product supply in Q2 2026 and YTD 2026 due to internal supply constraints.” The filing then adds the sentence the bulls need to read twice — Intel expects “these constraints to ease over the second half of 2026.”
That is a company telling you, in its own regulatory filing, that the shortage generating its pricing power is temporary and that it is actively working to end it. A 48% ASP increase driven by allocation is not the same asset as a 48% ASP increase driven by a technology lead. When supply normalises, the ASP tailwind reverses even if Intel ships more units. Anyone modelling Intel forward off a 59% DCAI growth rate without decomposing it into 48% price and 9% volume is building a forecast on a number that Intel has already told the SEC will not persist in the same form.
This is also where the competitive question bites. The live argument in data-centre silicon is performance per watt, fought against Arm-based server designs that have already displaced meaningful x86 socket volume at the hyperscalers. A 9% unit increase is not a share-gain number. The 10-Q attributes the ASP gain primarily to “a higher mix of premium products sold,” with demand-based pricing actions contributing to a lesser extent — Intel is monetising scarcity of its own premium SKUs. Real money today, a thinner moat than the headline suggests.
Intel Foundry: $293 million of external revenue against a $2.09 billion quarterly loss
Intel Foundry booked $5.77 billion of segment revenue in the quarter, but the 10-Q is careful to say that revenue “consists substantially of intersegment product and services revenue” — Intel selling wafers to Intel. The number that measures the actual foundry business, third-party foundry and assembly-and-test revenue from external customers, was $293 million in Q2 2026 and $467 million across the first half.
That $293 million cuts both ways. It is up more than thirteenfold from $22 million a year earlier, a genuine inflection. It is also 1.8% of consolidated revenue, and it sits against an Intel Foundry operating loss of $2.09 billion for the quarter — the segment loses roughly $7.13 for every $1 it earns from an outside customer. Annualised, Intel Foundry’s third-party business runs at about $1.2 billion against a market capitalisation north of $500 billion.
The significant offset is that the loss is narrowing fast: $2.09 billion against $3.17 billion a year earlier, a $1.08 billion improvement driven by yields, cycle times and factory scale. Zinsner put a hard number on the cost curve: “Intel Foundry has driven down the cost of our primary Panther Lake SKU by roughly 50% year-to-date.”
The 10-Q also settles a question that had hung over the stock for a year. Intel had left open whether it would complete Intel 14A at all without external commitments. The filing states that “during Q2 2026, we committed to completing development of Intel 14A, with a number of future Intel products designed to utilize the node and manufacturing expansion projects underway.” Intel 18A-P, the derivative node aimed at external customers, “entered into risk production” in June 2026 — risk production, note, not volume production. On the call, Tan said: “I’m pleased to see the increasing momentum on customer engagements for Intel 14A.” Momentum on engagements is not a signed anchor customer, and the market has been consistent about pricing that distinction. Our TSMC bull and bear analysis covers the incumbent Intel has to take that business from.
The $20 billion raise: what it fixes, what it costs
On 11 August Intel priced 210,526,315 shares at $95.00, upsized from an originally announced $15 billion deal, for net proceeds of approximately $19.7 billion. Underwriters hold a 30-day option on a further 31,578,947 shares. The stated use is general corporate purposes “including but not limited to capital expenditures and working capital.” FinanceFeeds covered the pricing overnight in Intel prices $20 billion stock sale at $95 a share.
Two details matter more than the headline number. First, the upsize: a company does not take a deal from $15 billion to $20 billion unless the book is oversubscribed by a wide margin, and CNBC reported demand well in excess of the deal size. Second, the price: $95 was a 2.6% discount to the 10 August close of $97.52 and 6.5% to the 7 August close of $101.65. Neither is a distressed print. This was a company selling equity into strength, not because it had to.
Now the dilution maths, which requires three numbers from two documents:
| Shares outstanding, 27 June 2026 | 5,043m | — |
| New shares issued at $95 | +210.5m | +4.17% |
| Underwriters’ 30-day option | +31.6m | +0.63% |
| Escrowed shares still owed to the US government | +143.0m | +2.84% |
| Fully loaded share count | 5,428.1m | +7.64% |
The 143 million escrowed shares are the piece that routinely gets left out. They are already committed, they will be issued as Secure Enclave milestones are met — Intel released 7 million in Q2 alone — and they dilute every existing holder just as surely as the underwritten offering does. Fully loaded, Intel’s share count is 7.6% higher than most screens still show.
What the raise buys is the strongest balance-sheet repair in Intel’s modern history. Pre-deal, Intel held $29.73 billion of cash and short-term investments against $50.54 billion of total debt — net debt of roughly $20.8 billion. Add $19.7 billion of net proceeds and net debt falls close to $1.1 billion. Intel goes into the 14A build essentially net-debt-free, which removes the financing overhang that has capped the multiple through the entire turnaround.
What it does not buy is time. Zinsner guided 2026 capex to “more than $20 billion” and said “we’re forecasting 2027 capital expenditures to be significantly above the 2026 levels.” Intel spent $6.19 billion on property, plant and equipment in the first half. The entire $19.7 billion raise is worth roughly one year of the capital plan Intel has just committed to — a landmark raise that funds about twelve months of the build.
The bull case: $148
Our bull case is $148, 51.5% above the 11 August close and about 4% above the 22 June closing peak of $140.94. It requires four things, and three of them are already in evidence.
One, the balance sheet is fixed, and it is fixed on terms that cost 4.2% of the company rather than a covenant package. Two, 14A is committed rather than conditional — the 10-Q language ends the “will they cancel it” debate that suppressed the multiple through 2025. Three, the foundry loss is compounding downward at roughly $1 billion a year with a 50% cost reduction on the lead product to point at. Four, and this is the one still outstanding, Intel names a marquee external 14A customer with volume attached.
Intel Products alone is annualising close to $19.3 billion of operating income at Q2’s run rate. Halve the foundry loss from about $8.4 billion annualised to roughly $4 billion, hold Products flat, and Intel generates well over $15 billion of annual operating income before any external foundry contribution. At $148 on a fully loaded 5.43 billion shares, the market capitalisation is roughly $803 billion — about 12 times a plausible FY2027 revenue line near $68 billion.
Be honest about what that multiple means. At $148 you are not paying a semiconductor multiple; you are paying a strategic-infrastructure multiple for the only leading-edge logic foundry on American soil, with the US government as a roughly 10% shareholder. Defensible, but not an earnings thesis, and investors should not pretend otherwise.
The bear case: $62
Our bear case is $62, 36.5% below spot. The number is not arbitrary. Intel fell from $140.94 on 22 June to $90.20 on 31 July — a 36.0% drawdown inside six weeks, with no accounting scandal, no guidance cut and no customer loss. One repeat of the drawdown this stock has already delivered this summer, from today’s price, lands at $62.5. The bear case is simply realised volatility applied twice.
The fundamental path is equally unexotic. Supply constraints ease exactly as Intel’s filing says they will, server ASPs give back part of a 48% increase, and DCAI decelerates hard because 9% volume growth cannot carry a 59% headline alone. Capex steps up “significantly” in 2027 against a foundry segment still losing north of $2 billion a quarter. No named 14A anchor customer materialises through the first half of 2027. Fully loaded dilution of 7.6% works against per-share numbers throughout.
Even at $62 Intel would carry a market capitalisation around $336 billion and would still be up roughly 68% from the 31 December 2025 close of $36.90. This is not a collapse thesis; it is a re-rating back toward the level the stock held before the April-to-June melt-up. For calibration, the analyst consensus target sat near $101 with a Hold rating across 32 covering analysts as of 10 August 2026, per Public.com’s compilation — the street is modelling almost no move at all.
What we are watching between now and mid-2027
Three concrete markers, each with a causal chain attached.
Q3 2026 results. Zinsner guided revenue of $15.8 billion to $16.8 billion with gross margin of 42%. The tell is not the revenue line; it is whether Intel repeats the “supply constraints easing” language and what happens to server ASPs when it does. If ASPs hold while volume accelerates, the bull case strengthens materially. If ASPs roll over as volume arrives, the 59% DCAI print was the peak.
A named 14A customer, and the contract structure behind it. Tan has committed to a 2028 high-volume ramp, and design commitments for a 2028 ramp get signed in 2026 and 2027, not 2028. Expect this to be the largest single driver of Intel’s multiple over the next four quarters, and expect the market to distinguish sharply between a second-source hedge and a lead-product commitment.
The escrow liability line in every future 10-Q. With 143 million shares still to release, this account will keep whipsawing GAAP results in the opposite direction to the share price. Read Intel’s operating income and ignore the net line. It is the only way to see the business.
Our view: the operating turnaround is real and better than the GAAP headline permits anyone to see, the balance sheet is now genuinely strong, and the valuation already prices a foundry outcome that $293 million of quarterly external revenue does not yet support. That is a legitimately two-sided setup, which is why the range is as wide as it is. For adjacent reading, our AMD bull and bear analysis and our Micron price prediction cover the two names most directly exposed to the same data-centre cycle.
Frequently asked questions
Why did Intel report an $11 billion loss on record revenue?
Because $12.5 billion of the loss was a non-cash, mark-to-market charge on shares Intel owes the US Department of Commerce under its August 2025 agreement. Intel carries the obligation as a derivative liability and revalues it quarterly. Intel’s stock rose sharply in the quarter, so the liability rose with it. Operating income was positive $1.80 billion.
How much did the $20 billion stock sale dilute Intel shareholders?
The 210.5 million new shares represent 4.17% dilution against the 5,043 million shares outstanding at 27 June 2026. Including the underwriters’ 31.6 million option and the 143 million escrowed shares still owed to the government, fully loaded dilution is approximately 7.64%.
Is Intel Foundry actually winning external customers?
Partially. External foundry and assembly-and-test revenue reached $293 million in Q2 2026, up from $22 million a year earlier — a thirteenfold increase, but still 1.8% of company revenue against a $2.09 billion segment operating loss. Intel has committed to completing Intel 14A and reports growing engagement, but has not disclosed a marquee anchor customer.
What is the difference between Intel 18A, 18A-P and 14A?
Intel 18A is the leading-edge node that entered high-volume production for Intel’s own products at the start of 2026. Intel 18A-P is the derivative aimed at external foundry customers; it entered risk production in June 2026. Intel 14A is the next generation, formally committed to completion in Q2 2026, with a high-volume ramp targeted for 2028.
What is Intel’s cash position after the raise?
Intel held $29.73 billion in cash and short-term investments against $50.54 billion of total debt at 27 June 2026. Adding roughly $19.7 billion of net proceeds takes net debt from about $20.8 billion to close to $1.1 billion, leaving the company approximately net-debt-neutral heading into its 2027 capital programme.
What would make the $148 bull case work?
Server ASPs holding as supply constraints ease, the Intel Foundry operating loss halving toward $4 billion annualised, and a named external 14A customer with volume commitments disclosed before the 2028 ramp. The first two are already trending correctly; the third is the binary event.
This article is analysis, not investment advice. Price targets are scenario estimates, not forecasts, and Intel’s share price can fall as well as rise.