INTC — Q2 2026 Financial Report Analysis

Fastest revenue growth in fifteen years, buried under a $12.5 billion accounting loss on the government's own stake

Intel's second quarter of 2026 (the three months ended June 27, 2026) produced two results that point in opposite directions, and the gap between them is the whole story.

The operating business turned. Revenue was $16.1 billion, up 25% from $12.9 billion a year earlier — what CEO Lip-Bu Tan called "our strongest revenue growth in more than fifteen years." Gross margin — the share of revenue left after the direct cost of making the chips — rose to 40.4% from 27.5%. Operating income, the profit from running the business before financing items and tax, swung to a positive $1,796 million from a $3,176 million loss.

Then, below the operating line, a single non-cash item wiped all of it out. "Interest and other, net" was negative $12,576 million, against negative $95 million a year ago, because Intel booked a $12.5 billion loss on the shares it has pledged to the U.S. government. The result: a net loss attributable to Intel of $11.0 billion, or $(2.16) per share, versus a $2.9 billion loss, or $(0.67), in Q2 2025.

That $12.5 billion charge got larger because Intel's share price went up. It is worth understanding why.

Headline figures

* Intel itself labels the year-over-year change on these lines "not meaningful," because both periods are losses. Segment revenue includes sales between Intel's own segments; intersegment eliminations were $(5,477)M in Q2 2026, which is why the segments sum to more than consolidated revenue. Source: Q2 2026 Form 10-Q and the July 23, 2026 earnings release (Exhibit 99.1).

The escrowed-shares charge: a loss that grows when the stock rises

Under the Warrant and Common Stock Agreement Intel signed with the U.S. Department of Commerce on August 22, 2025, Intel issued shares that sit in escrow and are released to the DOC as Intel performs under, and receives cash for, the CHIPS Act "Secure Enclave" program. Intel released 7 million of those shares during Q2 2026 and 13 million year to date, with 143 million still in escrow at quarter end.

Accounting treats that obligation as a derivative liability that has to be re-valued every quarter at the current share price. So when Intel's stock rises, the shares it owes the government become more valuable — and Intel books a loss for the difference. In Q2 2026 that loss was $12.5 billion ($13.6 billion year to date), which the 10-Q attributes explicitly to "an increase in our stock price." The liability on the balance sheet reached $15.6 billion at June 27, 2026, up from $2.7 billion at the end of 2025.

Nothing was paid out. No cash left the company. The charge is also why GAAP EPS of $(2.16) and management's non-GAAP EPS of $0.42 differ by $2.45 per share — the mark-to-market is the single largest reconciling item between the two. Intel also issued the DOC warrants on up to 241 million shares at $20.00, exercisable only if Intel ceases to own at least 51% of its foundry business; those are excluded from the share count because they are "neither currently nor expected to become exercisable."

The flip side is real dilution that has happened: weighted-average diluted shares were 5,104 million in Q2 2026 versus 4,369 million a year earlier, up 17%, reflecting the equity Intel has issued since mid-2025. Per-share results now get divided across a materially larger base.

Takeaway: Strip out the government-share mark-to-market and Intel's core business made money this quarter for the first time in years — $1.8 billion of operating income on 25% revenue growth. But the growth is priced, not shipped: client unit volumes fell 8% while average selling prices rose 27%, so the top line is being carried by a richer product mix and supply that cannot meet demand, not by selling more chips.

Where the growth came from — and what it really is

Data Center and AI (DCAI) was the standout: revenue $6,262 million, up 59%, with operating income of $2,474 million against $633 million a year ago. The 10-Q is specific about the driver: server revenue rose $2.0 billion on average selling price (ASP) increases of 48%, with "the majority of the increase in server ASPs... driven by a higher mix of premium products sold," and demand-based pricing actions contributing less. Server unit volume rose a comparatively modest 9%, "primarily driven by higher hyperscaler demand." Other DCAI revenue of $951 million was up $304 million on demand for purpose-built silicon (ASICs).

Client Computing and Physical AI Group (CCPG) — the renamed PC chip business — grew more slowly: revenue $8,877 million, up 13%, operating income $2,343 million (up $290 million). Here the mix effect is starker. Client revenue (notebook plus desktop) was $7.7 billion, up $1.1 billion, "primarily driven by ASP increases of 27%" — but volumes fell 8% year over year. Intel is selling fewer PC chips at much higher prices, partly because it is selling a premium-weighted mix and partly because it cannot make enough of them: "Market demand exceeded our available product supply in Q2 2026... due to industry-wide supply constraints." Management expects those client constraints "to ease over the second half of 2026."

CCPG's operating income grew far less than its revenue: $701 million of higher product profit and $258 million of lower operating expense were largely offset by $669 million of higher period charges, "primarily due to an inventory-related charge recognized in Q2 2026 to align product mix with customer demand." Writing down inventory to match what customers actually want is not a sign of a clean demand picture.

Intel Foundry — the manufacturing arm — had revenue of $5,765 million, up 31%, and narrowed its operating loss to $(2,089) million from $(3,168) million. Two caveats matter. First, almost all of that revenue is internal: with $5,477 million of intersegment eliminations consolidated away, Foundry is still overwhelmingly making chips for Intel's own product groups, not for external customers. Second, the loss narrowed mostly on the absence of prior-year charges — $1.4 billion of lower period charges, including the absence of $797 million of asset impairment and accelerated depreciation booked in Q2 2025 — while underlying product profit actually fell $340 million, "driven by an increased mix of higher-cost wafers manufactured on our Intel 18A process node." The newest, most advanced node is currently a margin drag, as new nodes usually are early in their ramp.

The same caution applies to consolidated gross profit, which rose $3.0 billion. Roughly half of that ($1.5 billion) is higher product profit; the other half is $1.5 billion of lower period charges, again largely the absence of last year's $797 million impairment. Underlying margin improvement is real but about half the size the headline suggests.

Two comparability points are worth holding onto. Intel deconsolidated Altera on September 12, 2025 after selling 51% of it, so Altera's revenue was inside Intel's numbers in Q2 2025 but shows up only as customer revenue now ($181 million in Q2 2026 versus $428 million contributed in Q2 2025) — this depresses the "All Other" segment, which fell 33% to $701 million, and flatters Foundry's growth rate, since Altera became an external foundry customer. And year-to-date results carry a $3.9 billion non-cash goodwill impairment taken in Q1 2026, substantially all related to Mobileye, which is why Intel is still $1,340 million in the red on operating income for the first half despite the strong second quarter.

The balance sheet: a $14.2 billion buyback of its own fab

The quarter's other large event was structural rather than operational. Intel bought out Apollo's 49% minority stake in the Ireland SCIP entity that owns Fab 34 for $14.2 billion in cash, taking full ownership of the fab, terminating the related operating agreements and extinguishing a $532 million liability tied to construction-delay damages. Because this was a purchase from a minority owner rather than an acquisition, $13.5 billion of the price went straight against equity (capital in excess of par) rather than through the income statement.

It was funded with cash, short-term investments and a $6.5 billion term loan, which Intel then repaid by issuing $6.5 billion of senior notes maturing between 2031 and 2066 at coupons from 4.65% to 6.20%. The consequences show up across the balance sheet:

• Cash and short-term investments: $29.7 billion, down from $37.4 billion at the end of 2025.
• Total debt: $50.5 billion, up from $46.6 billion.
• Total Intel stockholders' equity: $87.5 billion, down from $114.3 billion — the combined effect of the $14.8 billion year-to-date net loss and the $13.5 billion equity charge for the Apollo buyout.

Cash generation itself was strong: $7.0 billion from operations in the quarter (versus $2.1 billion) and $8.1 billion year to date (versus $2.9 billion), on higher revenue and lower operating expense. Gross capital expenditure fell 41% year over year to $2.65 billion. Management's own adjusted free cash flow measure was negative $8.4 billion for the quarter, but that is entirely the Apollo distribution — $12.2 billion of negative net partner contributions sits inside it. Before that one-off, the quarter's cash flows were comfortably positive.

Note the tension: capex is down 41% year over year, yet CFO Dave Zinsner said Intel is "meaningfully increasing our investments in equipment, clean room space, and substrates" to support expected growth. The spending increase is ahead, not behind — which means the free cash flow comparison gets harder from here.

Technology and risk

Intel says it began 2026 shipping its first products on Intel 18A in high-volume manufacturing, entered risk production on the derivative 18A-P node in June 2026, and during Q2 "committed to completing development of Intel 14A," its next-generation node, with expansion projects underway. Management is explicit that the pace of that expansion "will ultimately be dictated by the amount of committed demand for Intel 14A that we are able to obtain" — including from external customers still evaluating the node. That is the central unresolved question for the foundry strategy: 14A needs outside committed volume, and the filing does not claim to have it yet.

Two risks in the filing deserve flagging because they are not the usual boilerplate. Intel discloses that following the February 28, 2026 strikes on Iran and the ensuing conflict, Iran named Intel near the top of a published list of U.S. companies whose Middle East facilities it would target. Intel states that "a significant portion of our current and anticipated future revenues are generated from products on Intel 7 manufactured at our fabrication facility in Israel," that it is not insured for business interruption from war or political violence, and that its Israeli property is self-insured for such losses. Separately, Iranian strikes on Qatari energy fields have caused a global shortage of helium, a gas essential to semiconductor manufacturing.

Intel also faces a shareholder derivative suit filed in Delaware in March 2026 challenging the DOC share-and-warrant agreement itself; the plaintiff seeks to invalidate it, and defendants moved to dismiss in May 2026.

Outlook

Management guided Q3 2026 to revenue of $15.8–16.8 billion, GAAP gross margin of about 41.0% (42.0% non-GAAP), a 1% GAAP tax rate and GAAP diluted EPS of $0.31 ($0.38 non-GAAP), based on the midpoint of the revenue range. At the midpoint, $16.3 billion, that is roughly flat to slightly up sequentially — Intel is not guiding to continued 25% growth, and a positive GAAP EPS forecast implicitly assumes the escrowed-share revaluation does not swing violently again.

Intel expects "industry-wide shortages of substrates, memory and other critical components to persist into next year" and internal server supply constraints to continue, while client supply eases in the second half.

Our read: the operating turn is genuine but narrower than the headline. The clearest evidence of underlying improvement is the $4.97 billion year-over-year swing in operating income and the $7.0 billion operating cash flow — both hard to argue with. The qualifications are that roughly half the gross profit improvement is the absence of prior-year charges rather than better economics, that client chips are being sold in 8% lower volume, that Foundry's narrower loss also leans on absent charges while 18A wafers are currently dilutive to product profit, and that Foundry still sells almost entirely to Intel itself. The three things to watch in Q3 are whether DCAI's ASP-led growth holds once premium mix laps, whether easing client supply converts into unit growth rather than just lower prices, and whether Intel names a committed external customer for 14A. And GAAP earnings will stay hostage to the share price until the escrow unwinds: a strong stock quarter will keep producing large paper losses, and a weak one will produce paper gains. Read the operating line, not the bottom line.