[INTC] Intel: Q3 2026 earnings preview, can server CPU volume take over from a 48% price jump

Intel designs and manufactures its own x86 processors, selling PC chips to Dell, HP, Lenovo and other OEMs and Xeon server CPUs and custom silicon to cloud providers, with most wafers produced by its own Intel Foundry segment in the United States, Ireland and Israel[1]. The company is scheduled to report results for the third quarter of 2026 (fiscal quarter ending September 26, 2026) on 2026-10-22[2]. In the second quarter of 2026, the last one disclosed ahead of the Intel Q3 2026 earnings report, revenue rose 25% year over year to $16.1 billion, GAAP gross margin was 40.4% and non-GAAP gross margin 41.8%, non-GAAP earnings per share were $0.42, but a $12.5 billion fair-value loss on escrowed shares pushed the GAAP net loss attributable to Intel to $11.0 billion[3]. By segment, CCPG revenue was $8.88 billion, DCAI revenue was $6.26 billion, up 59% year over year, and Intel Foundry booked $5.77 billion of revenue against a $2.09 billion operating loss[4]. Management guided third-quarter revenue to $15.8-16.8 billion, non-GAAP gross margin to about 42% and non-GAAP EPS to about $0.38[5], raised 2026 capital expenditure guidance to more than $20 billion and held full-year non-GAAP operating expenses at about $16.5 billion[6]. Analyst consensus compiled by Drillr stands at $16.36 billion of third-quarter revenue from 23 analysts and $0.39 of EPS from 24 analysts[7], almost exactly the guidance midpoint; the August share sale added about 5% to the share count, and whether consensus EPS fully reflects that dilution cannot be verified[8].

Three things in this report deserve close attention. First is the quality of DCAI growth: the $2.0 billion increase in second-quarter server revenue came from a 48% rise in average selling prices while unit volume grew only 9% and demand exceeded supply[9], and with the consolidated guidance midpoint of $16.3 billion roughly flat with the second quarter and CCPG flagged as flat sequentially[10], whether DCAI can hold $6.26 billion of revenue and a 40% segment margin decides whether the guidance is met. Second is the composition of the Intel Foundry loss: the second-quarter loss narrowed by $1.08 billion year over year, but the improvement came from $1.4 billion of lower period charges while product profit actually fell by $340 million as higher-cost 18A wafers took a larger share of the mix[11], so whether the third-quarter 10-Q attributes the change to yield or to period charges will show whether the 18A ramp is narrowing the loss or reshaping it. Third is the cadence of cash consumption: total capital expenditure in the first half was about $7.6 billion, while full-year guidance above $20 billion implies more than $6 billion per quarter in the second half, and the $7.01 billion of second-quarter operating cash flow included working-capital release[12], so quarter-end cash, operating cash flow and capex read together will show whether the $23 billion raised is an advance reserve or is being steadily consumed.

Company Background and Business Structure

Intel was founded in Santa Clara, California in 1968, invented the x86 processor and is the only US company that both develops leading-edge logic process technology and manufactures it at scale. Its products are designed and sold by the Intel Products business, made up of the CCG segment (renamed the Client Computing and Physical AI Group, CCPG, in 2026) and the DCAI segment, while manufacturing is handled by the Intel Foundry segment and third-party suppliers; Intel Foundry also offers design and manufacturing services to external customers[1]. 2025 was a year of restructuring: new CEO Lip-Bu Tan carried out a restructuring plan that cut R&D and administrative expenses from $22.1 billion to $18.4 billion[13], and the US government accelerated $5.7 billion of CHIPS grants and $3.2 billion of Secure Enclave disbursements in exchange for up to 433 million shares and 241 million warrants[14]. Revenue in 2025 was $52.85 billion, flat with the prior year, with a GAAP operating loss of $2.21 billion and a net loss attributable to Intel of $267 million[13].

The three reportable segments differ greatly in size. In 2025, CCG generated $32.23 billion of revenue and $9.32 billion of operating income (a 29% margin), while DCAI generated $16.92 billion of revenue and $3.42 billion of operating income (a 20% margin)[15]; Intel Foundry reported $17.83 billion of segment revenue, of which $17.5 billion was intersegment sales to the product segments at prices intended to approximate market, with only $307 million from external customers, and an operating loss of $10.32 billion[16]. By the second quarter of 2026, CCPG had $8.88 billion of revenue at a 26% margin, DCAI had $6.26 billion at a 40% margin, Intel Foundry had $5.77 billion of revenue and a $2.09 billion loss, and consolidated revenue was $16.13 billion with GAAP operating income of $1.80 billion[4]. All Other includes Mobileye, which generated $507 million of revenue in the second quarter of 2026[17]; Intel holds a majority stake and recorded a $3.9 billion goodwill impairment in the first quarter[18].

Customers and the manufacturing network are both highly concentrated. The three largest customers accounted for 43% of 2025 revenue, with the largest at 19%[19]. On manufacturing, Intel 7 still supplied the majority of internal production in 2025 and is expected to supply almost half in 2026; Intel 3 is used for Xeon 6, Intel 18A entered high-volume manufacturing in late 2025 for Core Ultra Series 3, and the company expects 18A and its 18A-P derivative to be the key process for multiple generations of future client and server CPUs[20]. On capital structure, Arizona SCIP still receives contributions and distributions in proportion to Intel's and Brookfield's ownership, and such arrangements are expected to have an increasing effect on net income attributable to Intel as wafer volumes rise in Arizona and at Fab 34 in Ireland[21]; the 49% interest in Ireland SCIP was repurchased from Apollo on April 8, 2026 for $14.2 billion, and Intel now owns 100%[22].

Financial History and Current Position

The annual record is a curve of continuous decline followed by stabilization. Revenue was $54.23 billion in 2023 with operating income of $93 million; 2024 revenue was $53.10 billion with a GAAP operating loss of $11.68 billion; 2025 revenue was $52.85 billion, flat with the prior year, gross margin was 34.8%, R&D and administrative expenses fell from $22.1 billion to $18.4 billion, the GAAP operating loss was $2.21 billion, the net loss attributable to Intel was $267 million and non-GAAP EPS was $0.42[13]. Operating cash flow in 2025 was $9.70 billion, capital expenditure within investing activities was $14.65 billion (versus $23.94 billion in 2024) with another $3.03 billion within financing activities; the gap was filled by $12.71 billion of US government equity proceeds, $5.11 billion of partner contributions and $6.16 billion from the Altera sale[23].

The two quarters of 2026 already show a different picture. First-quarter revenue was $13.58 billion, non-GAAP gross margin 41.0%, non-GAAP EPS $0.29, the GAAP operating loss $3.14 billion and the net loss attributable to Intel $3.73 billion, with operating cash flow of $1.10 billion, gross capital expenditure of $4.96 billion and adjusted free cash flow of -$2.02 billion[24]. Second-quarter revenue rose 25% year over year to $16.13 billion, GAAP gross margin was 40.4% and non-GAAP 41.8%, GAAP operating income was $1.80 billion and non-GAAP $2.77 billion, non-GAAP net income attributable to Intel was $2.20 billion or $0.42 per share; the GAAP net loss attributable to Intel was $11.0 billion or -$2.16 per share, with the difference almost entirely the $12.53 billion fair-value loss on escrowed shares[3].

The change on the cash side is larger. Second-quarter operating cash flow was $7.01 billion and gross capital expenditure $2.65 billion, but after $12.22 billion of net partner distributions (mainly the Ireland SCIP repurchase payment to Apollo), adjusted free cash flow was -$8.42 billion[12]. At June 27, cash and equivalents were $12.87 billion and short-term investments $16.85 billion, together $29.73 billion; long-term debt was $48.55 billion and short-term debt $1.99 billion, and Intel stockholders' equity was $87.54 billion[25]. In August the company sold 210.5 million shares at $95 per share, the underwriters then exercised in full their option for another 31.58 million shares, the offering was upsized from $15 billion to $20 billion, net proceeds before the option were about $19.7 billion, and the stated use is general corporate purposes including capital expenditures and working capital[8].

Operating Model

Revenue is almost a simple function of units times price. CCPG revenue equals client (notebook plus desktop) units times client average selling price plus other CCPG revenue, which were $7.7 billion and $1.1 billion in the second quarter; DCAI revenue equals server units times server ASP plus other revenue such as ASICs and networking, with other DCAI revenue of $951 million in the quarter[9]. Of Intel Foundry's $5.77 billion segment revenue, $5.5 billion was intersegment sales to the product segments at prices intended to approximate market and is eliminated entirely on consolidation, while external revenue of $293 million came mainly from Altera after its deconsolidation[26]. Consolidated revenue grew 25% year over year in the second quarter, and almost all of that growth came from price: server ASPs rose 48% and client ASPs 27%, while units were +9% and -8% respectively[9].

Operating income is the sum of two profitable product segments and one loss-making manufacturing segment. In the second quarter, CCPG operating income of $2.34 billion plus DCAI operating income of $2.47 billion made $4.82 billion (a 32% margin); adding Intel Foundry's -$2.09 billion, All Other's +$230 million, corporate unallocated expenses of -$1.42 billion and intersegment eliminations of +$254 million gives consolidated operating income of $1.80 billion[4]. Product-segment profit is highly sensitive to price, but the unit cost of premium products is rising too: CCPG product profit rose $701 million in the quarter but was largely offset by a $669 million inventory-related charge, and DCAI product profit rose $1.7 billion but was likewise partly offset by higher unit costs[27]. Intel Foundry's loss consists of depreciation (the substantial majority of consolidated depreciation sits in this segment), process R&D and high-cost wafers on early nodes; the loss narrowed by $1.08 billion year over year because period charges fell by $1.4 billion, while a higher mix of high-cost 18A wafers cut product profit by $340 million[11]. Net income is additionally driven by non-operating items: the $12.5 billion second-quarter fair-value loss on escrowed shares grew with the share price, and the related liability rose to $15.6 billion[28].

The cash formula is adjusted free cash flow = operating cash flow - gross capital expenditure + capital-related government incentives +/- partner contributions or distributions - finance lease payments. In the second quarter those items were $7.01 billion, -$2.65 billion, $60 million, -$12.22 billion and -$617 million, giving -$8.42 billion[12]. Total capital expenditure in the first half was about $7.6 billion ($6.19 billion within investing plus $1.42 billion within financing)[29], while full-year guidance is above $20 billion with 2027 meaningfully higher[6], which means second-half spending must accelerate sharply. The funding sources, in order, are $29.7 billion of quarter-end cash and short-term investments, the $20 billion August offering (about $23 billion including the over-allotment), Arizona SCIP partner contributions and $678 million of AMIC investment tax credits in the first half[30].

Industry and Competitive Position

In client CPUs, Intel shares the x86 ecosystem with AMD and faces the ARM camp of Apple's M series, Qualcomm's Snapdragon and MediaTek's Kompanio, and the 10-K expects competition to intensify further in 2026[31]. In the data center, AI workloads have sent most incremental compute budgets to NVIDIA's GPU systems, AMD competes head-on in x86 servers, hyperscalers such as Amazon, Google, Meta and Microsoft develop their own silicon, and Broadcom competes in custom ASICs[32]. Management acknowledges that it still lags AMD on some CPU architecture metrics but frames the near-term challenge as growing supply rather than winning share, says it secured multiple long-term customer agreements in the second quarter, and describes Xeon 6 as one of the fastest-ramping products in Intel's history[10].

In foundry, Intel considers itself one of only three companies alongside TSMC and Samsung investing in 2nm-class technology, but admits limited experience in the third-party foundry business and notes that many of the largest potential customers are fabless companies whose products compete with its own[33]. The 10-K states that Intel has been unsuccessful to date in securing any significant external foundry customer for any node, that prospects for a significant 14A customer are uncertain, and that without one it may pause or discontinue development of 14A and subsequent nodes[34]. Industry-wide shortages of substrates, memory and leading-edge logic constrain Intel and its competitors alike: market demand exceeded available supply in the second quarter, and the company expects these shortages to persist into next year[35]. The comparison has limits, because Intel discloses only year-over-year percentages for ASPs and units rather than absolute values, and Intel Foundry's segment revenue is intersegment sales at transfer prices, so its loss cannot be compared directly with an external foundry.

Core Debates

Was the 59% jump in server CPU revenue a price spike under shortage, or the start of a volume ramp?

DCAI was the source of the second-quarter profit jump, so the quality of its growth decides whether guidance can be met. Second-quarter DCAI revenue was $6.26 billion with operating income of $2.47 billion, and the margin rose from 16% a year earlier to 40%[4]. The 10-Q is explicit, however, that the $2.0 billion increase in server revenue came from a 48% rise in average selling prices, attributed mainly to a higher mix of premium products and secondarily to demand-based pricing to offset higher input costs, while units grew only 9% and market demand exceeded internal supply[9]. Price increases are easiest under shortage and easiest to lose once supply recovers. The consolidated guidance midpoint of $16.3 billion is almost flat with the second quarter[5], and management flagged CCPG as flat sequentially[10], so whether DCAI holds $6.3 billion decides whether the guidance is achieved.

Both readings find support in current disclosure. Management expects supply growth to be weighted toward the end of the third quarter and into the fourth, sees server CPU industry unit demand growing at a strong double-digit rate in 2026 and 2027, and secured multiple long-term customer agreements in the second quarter[6]; the first-quarter call explained that these agreements typically carry volume and pricing terms over 3 to 5 years, with Google among them[36]. The benign reading is that the Xeon 6 premium mix and long-term agreements lock in price, volume takes over once supply is released, and revenue and margin step up together. The alternative reading is that a meaningful part of the ASP increase is a shortage premium: server ASPs fell 4% for full-year 2025, but prices had already begun rising in the second half on higher demand[37], so the third-quarter comparison base is higher, and if the ASP increase narrows while units remain capacity-constrained, DCAI revenue would stall in the low $6 billion range and the 40% margin would be hard to sustain. The 10-K also warns that competition from AMD, NVIDIA and hyperscaler-designed ARM chips continues to intensify[32]. In financial terms, server ASP and units jointly determine DCAI revenue; ASP gains fall almost entirely to segment profit, while unit growth brings costs that scale with wafer capacity.

Three numbers matter in the third quarter. One is whether DCAI revenue is no lower than $6.26 billion and the 10-Q shows server unit growth back above 9% year over year; two is how far the server ASP increase narrows and whether the 10-Q attributes it to mix or to pricing actions; three is whether the DCAI segment margin holds 35% and whether the company again says demand exceeds supply. The falsifiers are equally specific: if the ASP increase falls below 20% while unit growth stays in single digits, the second quarter was a one-time shortage premium; if the 10-Q mentions pricing actions taken in response to competition, or the segment margin drops below 30%, the benign reading does not hold.

The PC market is shrinking; how long can Intel's client business hold up on price alone?

CCPG is still Intel's largest segment, contributing $32.23 billion of revenue and $9.32 billion of operating income in 2025[15]. Second-quarter client revenue of $7.7 billion was $1.1 billion higher year over year, entirely from a 27% rise in ASPs, while units fell 8%[9]. Management at the same time expects 2026 PC industry demand to fall by low double digits year over year, hurt by rising memory prices and supply constraints, with third-quarter CCPG revenue roughly flat sequentially as edge AI growth offsets softness in core PCs[6]. This is the first quarter that tests whether price increases can persist through falling demand.

The existing evidence shows both mix upgrade and signs of misjudgment. Second-quarter CCPG revenue was $8.88 billion with operating income of $2.34 billion and a 26% margin[4], of which AI PCs made up 67% of revenue and edge about 10%[6]. The 10-Q attributes the ASP increase primarily to a higher premium mix and secondarily to selective price increases to offset input costs, attributes the unit decline to industry-wide supply constraints, and expects those constraints to ease in the second half[9]. In the same quarter, however, the segment recorded a $669 million inventory-related charge to align product mix with customer demand, nearly offsetting a $701 million increase in product profit, so margin did not keep pace with revenue[27]. Management said on the call that excess client capacity can be redirected to servers[10]. The benign reading is that Panther Lake and AI PCs have permanently moved the mix upmarket, the company prioritized higher-margin products under tight supply, and the unit decline was a deliberate choice; the alternative reading is that part of the ASP increase reflects OEMs pulling orders forward and accepting price rises under shortage expectations, so that once memory inflation compresses system shipments and channel restocking is complete, unit declines widen and ASPs fall back, while the inventory charge shows the company's mix judgment has already drifted. In financial terms, client revenue equals units times ASP, and higher ASPs also raise unit cost, so segment margin is less elastic than revenue.

The third-quarter watch points are the revenue range, the price-volume split and management's framing. One is whether CCPG revenue lands in the flat range of $8.8-9.3 billion and whether the 10-Q again mentions inventory-related charges; two is whether the client ASP increase stays above 15% and the unit decline narrows to within 8%; three is whether management's view of 2026 PC industry demand has changed and whether the edge and physical AI share of revenue keeps rising. If the unit decline widens beyond 15% while the ASP increase narrows below 10%, or another large inventory-related charge appears or the segment margin falls below 22%, the price-supported logic is falsified.

As 18A output ramps, is Intel Foundry's loss really narrowing, or just changing shape?

Intel Foundry is the entire source of Intel's losses and the destination of the newly raised capital. The segment lost $10.32 billion in 2025[16] and still lost $2.09 billion in the second quarter, while the two product segments earned $4.82 billion in the same quarter[4]. The company raised 2026 capital expenditure to more than $20 billion with 2027 higher still[6], then sold $20 billion of stock in August[8], and almost all of that money is going into this segment. Whether the loss narrows depends on whether 18A yields can outrun the high-cost wafers of the ramp, and whether 14A wins an external customer decides whether this investment has a second revenue stream.

The second-quarter improvement does not survive decomposition. Intel Foundry revenue was $5.77 billion and the loss was $2.09 billion, $348 million less than the first quarter[6] and $1.08 billion less than a year earlier; but the 10-Q shows the $1.4 billion year-over-year improvement came from lower period charges (the prior-year quarter carried $797 million of asset impairment and accelerated depreciation), while product profit actually fell by $340 million because a higher mix of high-cost 18A wafers offset the cost reductions on Intel 3 and Intel 4[11]. On the call, management said 18A output rose more than 50% quarter over quarter and 25% above internal targets with yields ahead of expectations; 18A-P has entered risk production, 14A is ahead of 18A's schedule on defect density and transistor performance, PDK 0.5 is complete and PDK 0.9 is on track for October, with internal 14A products in risk production in the second half of 2027 and volume production in 2028[38]. External revenue of $293 million came mainly from Altera after its deconsolidation[26]. The 10-K wording is that no node has secured a significant external customer to date[34]; the second-quarter 10-Q changes to say the company has committed to completing 14A development, but the scale and pace of expansion will be dictated by committed demand from the internal roadmap and design wins with potential significant external customers[30]. The optimistic reading is that the yield curve has passed its steepest section, product profit turns positive from the third quarter and the loss narrows faster; the cautious reading is that the 18A share is still rising quickly, every additional high-cost wafer eats the yield benefit, the loss will not narrow until 2027, and 14A customers will not sign before PDK 1.0. In financial terms, Intel Foundry's loss equals revenue at internal transfer prices minus depreciation, process R&D and wafer cost, the loss flows directly into consolidated operating income, and external customer commitments determine 14A expansion and post-2027 external revenue.

The third quarter comes down to the attribution of the loss, the 14A milestone and consolidated gross margin. One is whether the Intel Foundry loss narrows to within $1.9 billion and whether the 10-Q attributes the change to yield and unit cost or to period charges; two is whether 14A PDK 0.9 is delivered in October as planned and whether the company's language on external customer commitments shifts from "dictated by" to "secured"; three is whether consolidated non-GAAP gross margin reaches the guided 42%, versus 41.8% in the second quarter[3], and whether management again flags 18A product cost as a drag on gross margin. If the loss widens beyond $2.4 billion and is attributed to the high-cost 18A wafer mix, or PDK 0.9 slips or the company repeats that it has not secured a significant external customer, the optimistic reading is falsified.

After a $20 billion-plus capex plan and a $23 billion share sale, how long does Intel's cash last?

The pace of capital spending was stepped up sharply in the second quarter. On the second-quarter call the company raised 2026 capital expenditure from "flat with 2025" at the start of the year to more than $20 billion and said 2027 would be meaningfully higher, with the vast majority directed to its US manufacturing network[6]; on the first-quarter call the framing had still been flat for the year with equipment spending rising[36]. Two weeks later the company sold stock at $95 per share, about 242 million shares including the over-allotment, in a $20 billion offering[8]. Total capital expenditure in the first half was only about $7.6 billion[29], so full-year guidance implies more than $6 billion per quarter in the second half. Meanwhile the company paid $14.2 billion in April to repurchase Apollo's interest in Ireland SCIP, bridged with a $6.5 billion term loan that was then refinanced with fixed-rate notes of the same size[22], and at the end of June cash and short-term investments had fallen to $29.73 billion with long-term debt of $48.55 billion[25]. The third quarter is the first with the offering proceeds in hand, and it will show whether capex really accelerates and whether operating cash flow keeps up.

The quality of operating cash flow is the crux of this debate. Second-quarter operating cash flow was $7.01 billion, gross capital expenditure $2.65 billion and adjusted free cash flow -$8.42 billion; the first-quarter figures were $1.10 billion, $4.96 billion and -$2.02 billion[12]. Full-year 2025 operating cash flow was $9.70 billion against $14.65 billion of capital expenditure within investing activities, with the gap filled by $12.71 billion of government equity proceeds, $5.11 billion of partner contributions and $6.16 billion from the Altera sale[23]. Management said the increased capex is mostly tooling for existing high-volume nodes (Intel 3, 18A, 18A-P), skewed to more expensive front-end capacity, and is committed only after yield, performance and customer demand are validated; the timing of AMIC investment tax credit receipts lags[10]. The offering 8-K states the use of proceeds as general corporate purposes, including capital expenditures and working capital, and stresses maintaining an investment-grade rating[8]. One reading is that, with product-segment margins having jumped, operating cash flow can already cover most of a $20 billion capex budget and the offering is an advance reserve for larger 2027 expansion; the other is that second-quarter operating cash flow included working-capital release, and if the third quarter falls back to $4-5 billion while capex jumps to $6 billion as guided, the proceeds will be steadily consumed before the 2027 expansion, with Arizona SCIP distributions and Fab 34 output commitments still ahead, the latter carrying volume-related damages if minimum purchases are missed from completion or the third quarter of 2027[21]. In financial terms, gross capital expenditure directly reduces adjusted free cash flow and the cash balance, and when capex accelerates without matching operating cash flow growth, future depreciation rises and Intel Foundry's fixed costs climb further.

Three numbers matter in the third quarter: capex, operating cash flow and quarter-end cash. One is whether gross capital expenditure rises into the $4.5-6.5 billion range and the company keeps full-year guidance above $20 billion; two is whether operating cash flow holds above $5 billion and adjusted free cash flow returns to within -$1 billion; three is whether quarter-end cash and short-term investments reach $45 billion or more after the offering, and whether there is any new debt issuance or SCIP-related outflow. If operating cash flow falls below $4 billion while capex accelerates as guided, or the company cuts capex guidance or delays the 2027 expansion, the current understanding has to be rewritten.

Risks and Falsifiers

Industry-wide supply shortage is the first risk, and one the company cannot control. Shortages of substrates, memory and leading-edge logic are expected to persist into 2027; they limit Intel's own shipments (market demand exceeded available supply in the second quarter[35], and management has said it can meet only half of CPU demand) and also suppress PC system demand through memory inflation; the company mitigates through long-term supply agreements and capacity reallocation[10]. The exposed lines are CCPG and DCAI, which together produced $15.1 billion of revenue and $4.8 billion of operating income in the second quarter; every $1 billion of revenue lost to supply is worth about $300 million of profit at the 32% segment margin[4]. If the third-quarter 10-Q stops attributing unit declines to supply constraints, or management says supply growth was released at quarter-end as planned, this risk is receding.

US government ownership and escrowed shares bring non-operating volatility and governance constraints. The fair-value loss on escrowed shares grows as the share price rises: $12.5 billion in the second quarter, a liability balance of $15.6 billion and a cumulative first-half loss of $13.6 billion, all non-cash; if Intel's ownership of the foundry business falls below 51%, the government may exercise 241 million warrants at $20 per share[28]. The agreement could also be modified unilaterally by future legislation or administrative action[14]. The warrants correspond to about 5% of the share count at an exercise price far below market. If the third-quarter 10-Q adds no new contingency related to the government agreement and the company keeps 100% ownership of the foundry business, this risk stays within the range of book-value swings.

The shortage premium inside server pricing is the third risk. The 48% server ASP increase contains a shortage premium; as the comparison base rises with the second-half 2025 price recovery[37] and competition from AMD and hyperscaler ARM chips intensifies[32], the increase could narrow from the third quarter while internal wafer supply still limits units, so DCAI revenue and margin fall together. The exposed line is DCAI, with second-quarter revenue of $6.26 billion and operating income of $2.47 billion, half of Intel Products profit[4]; at second-quarter scale, every 10 percentage points less of server ASP increase is worth about $400-500 million of quarterly revenue, almost all of it profit. The falsifying observation is third-quarter DCAI revenue of at least $6.26 billion, server unit growth of at least 9% year over year and a segment margin of at least 35%.

Falling PC demand combined with memory inflation is the fourth risk. PC industry demand is down by low double digits[6], OEMs may cut orders and resist price increases, client unit declines widen and ASP gains fade, while inventory built for a premium mix needs further charges. The exposed line is CCPG, with second-quarter revenue of $8.88 billion and operating income of $2.34 billion[4]; a single inventory-related charge in the second quarter was $669 million, equal to 29% of segment profit[27]. The falsifying observation is third-quarter CCPG revenue of at least $8.8 billion, a client ASP increase of at least 15% and no new inventory-related charge in the 10-Q.

18A yield and 14A customers are the fifth risk, and the source of potential asset impairment. If 18A yield improvement fails to keep pace with the output ramp, a rising share of high-cost wafers keeps depressing Intel Foundry product profit[11]; if 14A does not win a significant external customer, the company may pause subsequent node development under the 10-K's stated contingency[34]. The exposed line is Intel Foundry, with a 2025 loss of $10.32 billion[16] and a second-quarter loss of $2.09 billion; the company holds $105.7 billion of net property, plant and equipment[25], the great majority in the foundry business, and pausing 14A would bring a major impairment. The falsifying observation is a third-quarter segment loss of no more than $1.9 billion attributed by the 10-Q to yield improvement, PDK 0.9 delivered on time and no renewed mention of pausing 14A.

The funding gap is the sixth risk. Capital expenditure accelerates as guided to more than $6 billion per quarter in the second half, but if operating cash flow falls back to $4-5 billion, the offering proceeds are consumed before the larger 2027 expansion and the company must borrow or issue stock again, raising dilution and interest burden. The exposed lines are 2026 capital expenditure above $20 billion with 2027 higher[6], cash and short-term investments of $29.73 billion and long-term debt of $48.55 billion at the end of June[25], and the August sale of about 242 million shares that already diluted the share count by about 5%[8]. The falsifying observation is third-quarter operating cash flow of at least $5 billion, quarter-end cash and short-term investments of at least $45 billion and no new financing announced.

What to Watch Next

• Server CPU price versus volume: DCAI revenue was $6.26 billion in the second quarter of 2026 with server ASP +48%, units +9% and a 40% segment margin. Watch whether third-quarter revenue holds at or above $6.26 billion, how the 10-Q splits and attributes price and volume, and whether the margin holds 35%. A drop below $6 billion with a narrowing ASP gain, an ASP increase below 20% with single-digit unit growth, a margin below 30% or pricing actions taken in response to competition would falsify the volume-ramp reading.
• Client price versus PC units: CCPG revenue was $8.88 billion with client ASP +27%, units -8% and a 26% margin. Watch whether revenue lands in the flat $8.8-9.3 billion range, whether the ASP increase stays at or above 15% and the unit decline within 8%, and whether inventory charges recur. Revenue below $8.8 billion, a unit decline beyond 15% with an ASP gain below 10%, another inventory charge or a margin below 22% would falsify the price-supported reading.
• Intel Foundry loss and the 18A ramp: the segment lost $2.09 billion in the second quarter, PDK 0.5 is complete, PDK 0.9 is due in October with no significant external customer, and consolidated non-GAAP gross margin was 41.8% against 42% guided. Watch whether the loss narrows to within $1.9 billion and is attributed to yield rather than period charges, whether PDK 0.9 ships on time and the customer language shifts to "secured", and whether gross margin reaches 42%. A loss beyond $2.4 billion attributed to high-cost 18A wafers, a PDK 0.9 delay, a repeated statement that no significant external customer has been secured, or renewed warnings that 18A cost is depressing gross margin would falsify the optimistic reading.
• Capex acceleration and cash runway: second-quarter gross capital expenditure was $2.65 billion, operating cash flow $7.01 billion, adjusted free cash flow -$8.42 billion and quarter-end cash and short-term investments $29.73 billion. Watch whether capex rises to $4.5-6.5 billion with full-year guidance maintained, whether operating cash flow holds at or above $5 billion and free cash flow returns to within -$1 billion, and whether cash and short-term investments reach $45 billion or more with no new financing. Operating cash flow below $4 billion with capex accelerating, a cut to capex guidance, a delayed 2027 expansion or new debt or equity financing would falsify the current reading.

Conclusion

Intel's business consists of two profitable product segments and one loss-making manufacturing segment, and the turn in the second quarter of 2026 was driven almost entirely by price. Consolidated revenue rose 25% year over year to $16.13 billion and non-GAAP EPS was $0.42[3], with server ASPs up 48% and client ASPs up 27% while units were only +9% and -8%[9]; CCPG and DCAI together earned $4.82 billion and Intel Foundry lost $2.09 billion[4]. On the cash side, $29.73 billion of cash and short-term investments at the end of June[25] plus the $20 billion August offering[8] stand against full-year capital expenditure above $20 billion and a higher 2027[6]. The unresolved central relationships are whether volume can take over from price to support DCAI's 40% margin once supply is released, whether 18A yields can outrun the rising share of high-cost wafers, and whether operating cash flow can keep pace with accelerating capex.

The two independent interpretations published after the second quarter map onto the server debate and the foundry debate respectively. Mwangi Enos of Yahoo Finance relayed Goldman Sachs's response on July 25: second-quarter revenue of $16.1 billion beat Goldman's own $14.3 billion and the Street's $14.4 billion, gross margin of 41.8% beat 39.3%, and non-GAAP EPS of $0.42 was nearly double Goldman's $0.23 estimate, so Goldman raised its estimates by 49% on average, acknowledged that Intel benefits from server demand growth and holds genuine geopolitical tailwinds as a US foundry champion, but judged that AMD, NVIDIA and Broadcom offer better revenue visibility and risk/reward and kept a neutral stance[39]. Rohail Saleem of Wccftech relayed a Piper Sandler report on September 18: Amazon, Apple, AMD, Google, Tesla, Microsoft, NVIDIA and Qualcomm are all evaluating 14A, defect density reached 0.5 in June with a target of 0.1 to 0.2 by the first quarter of 2027, and CFO Zinsner called 14A's rate of improvement the fastest since the 22nm era of 2012; the author judged that TSMC capacity is tight through 2028 and the industry is looking for an alternative, while also noting that Lip-Bu Tan has repeatedly said he will approve large-scale expansion only after locking in 14A customers[40]. The two views do not conflict but sit on different time scales: Goldman accepts the demand signal as real yet cites limited visibility as its reason not to raise its view, which is exactly why "volume or price in third-quarter DCAI" is the test; Piper Sandler's evaluation list is long, but the article itself concedes that none has signed, which is consistent with the 10-K's wording, and its 14A timeline (risk production in the second quarter of 2027, volume in the first quarter of 2028) differs slightly from the company's call, so the company's disclosure is taken as the reference. Neither piece discusses the effect of 18A cost on gross margin or evaluates the Intel Foundry loss.

What would materially change the current understanding is a combination of numbers rather than any one of them. If third-quarter DCAI revenue is at least $6.26 billion with server unit growth back above 9%, CCPG lands in the $8.8-9.3 billion range with no further inventory charge, the Intel Foundry loss narrows to within $1.9 billion and is attributed to yield, and operating cash flow is at least $5 billion with quarter-end cash and short-term investments of $45 billion or more, the "price first, volume next" understanding would be substantially strengthened. Conversely, if the server ASP increase drops below 20% while units still grow only in single digits, the client unit decline widens beyond 15%, the Intel Foundry loss widens beyond $2.4 billion on high-cost 18A wafers, and operating cash flow falls below $4 billion while capex still accelerates as guided, then the second quarter looks more like a shortage premium and the current understanding needs to be rewritten.

Sources

[1] INTC 10-K filed 2026-01-23 · business overview and segments · 2026-01-23 · 10-K · https://www.sec.gov/Archives/edgar/data/50863/000005086326000011/

[2] Drillr earning_call_calendar (updated 2026-09-21) · INTC earnings call 2026-10-22 · 2026-09-21 · Drillr earning_call_calendar · https://gateway.drillr.ai/mcp/private

[3] INTC 8-K filed 2026-07-23 · Q2 2026 results summary · 2026-07-23 · 8-K · https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0000050863&type=8-K&dateb=&owner=include&count=40

[4] INTC 10-Q filed 2026-07-24 · Q2 2026 segment results table · 2026-07-24 · 10-Q · https://www.sec.gov/Archives/edgar/data/50863/000005086326000157/

[5] INTC 8-K filed 2026-07-23 · Q3 2026 business outlook · 2026-07-23 · 8-K · https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0000050863&type=8-K&dateb=&owner=include&count=40

[6] INTC 2Q26 earnings call 2026-07-23 · Q2 2026 guidance and capital plan · 2026-07-23 · earnings-call · https://www.intc.com/events-and-presentations

[7] Drillr analyst_financial_estimates (updated 2026-09-21) · 3Q26 revenue and EPS estimates · 2026-09-21 · Drillr analyst_financial_estimates · https://gateway.drillr.ai/mcp/private

[8] INTC 8-K filed 2026-08-12 · $20 billion common stock offering · 2026-08-12 · 8-K · https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0000050863&type=8-K&dateb=&owner=include&count=40

[9] INTC 10-Q filed 2026-07-24 · Q2 2026 CCPG and DCAI revenue drivers · 2026-07-24 · 10-Q · https://www.sec.gov/Archives/edgar/data/50863/000005086326000157/

[10] INTC 2Q26 earnings call 2026-07-23 · Q2 2026 risks and Q&A · 2026-07-23 · earnings-call · https://www.intc.com/events-and-presentations

[11] INTC 10-Q filed 2026-07-24 · Q2 2026 Intel Foundry operating loss drivers · 2026-07-24 · 10-Q · https://www.sec.gov/Archives/edgar/data/50863/000005086326000157/

[12] INTC 8-K filed 2026-07-23 · Q2 2026 cash flow and adjusted free cash flow · 2026-07-23 · 8-K · https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0000050863&type=8-K&dateb=&owner=include&count=40

[13] INTC 10-K filed 2026-01-23 · FY2025 consolidated statement of operations · 2026-01-23 · 10-K · https://www.sec.gov/Archives/edgar/data/50863/000005086326000011/

[14] INTC 10-K filed 2026-01-23 · U.S. Government Agreement · 2026-01-23 · 10-K · https://www.sec.gov/Archives/edgar/data/50863/000005086326000011/

[15] INTC 10-K filed 2026-01-23 · FY2025 Intel Products segment results · 2026-01-23 · 10-K · https://www.sec.gov/Archives/edgar/data/50863/000005086326000011/

[16] INTC 10-K filed 2026-01-23 · FY2025 Intel Foundry revenue, external revenue and operating loss · 2026-01-23 · 10-K · https://www.sec.gov/Archives/edgar/data/50863/000005086326000011/

[17] INTC 10-Q filed 2026-07-24 · All Other and Mobileye · 2026-07-24 · 10-Q · https://www.sec.gov/Archives/edgar/data/50863/000005086326000157/

[18] INTC 10-Q filed 2026-04-24 · Q1 2026 Mobileye goodwill impairment · 2026-04-24 · 10-Q · https://www.sec.gov/Archives/edgar/data/50863/000005086326000079/

[19] INTC 10-K filed 2026-01-23 · customer concentration · 2026-01-23 · 10-K · https://www.sec.gov/Archives/edgar/data/50863/000005086326000011/

[20] INTC 10-K filed 2026-01-23 · process technology roadmap and node mix · 2026-01-23 · 10-K · https://www.sec.gov/Archives/edgar/data/50863/000005086326000011/

[21] INTC 10-K filed 2026-01-23 · Semiconductor Co-Investment Program and Fab 34 damages · 2026-01-23 · 10-K · https://www.sec.gov/Archives/edgar/data/50863/000005086326000011/

[22] INTC 10-Q filed 2026-07-24 · Ireland SCIP repurchase completed · 2026-07-24 · 10-Q · https://www.sec.gov/Archives/edgar/data/50863/000005086326000157/

[23] INTC 10-K filed 2026-01-23 · FY2025 consolidated cash flow statement · 2026-01-23 · 10-K · https://www.sec.gov/Archives/edgar/data/50863/000005086326000011/

[24] INTC 8-K filed 2026-04-23 · Q1 2026 results and non-GAAP reconciliation · 2026-04-23 · 8-K · https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0000050863&type=8-K&dateb=&owner=include&count=40

[25] INTC 10-Q filed 2026-07-24 · balance sheet at June 27, 2026 · 2026-07-24 · 10-Q · https://www.sec.gov/Archives/edgar/data/50863/000005086326000157/

[26] INTC 10-Q filed 2026-07-24 · Q2 2026 Intel Foundry revenue and external revenue · 2026-07-24 · 10-Q · https://www.sec.gov/Archives/edgar/data/50863/000005086326000157/

[27] INTC 10-Q filed 2026-07-24 · Q2 2026 Intel Products operating income drivers · 2026-07-24 · 10-Q · https://www.sec.gov/Archives/edgar/data/50863/000005086326000157/

[28] INTC 10-Q filed 2026-07-24 · Escrowed Shares loss and derivative liability · 2026-07-24 · 10-Q · https://www.sec.gov/Archives/edgar/data/50863/000005086326000157/

[29] INTC 10-Q filed 2026-07-24 · six-month 2026 cash flow statement · 2026-07-24 · 10-Q · https://www.sec.gov/Archives/edgar/data/50863/000005086326000157/

[30] INTC 10-Q filed 2026-07-24 · future node development and manufacturing expansion · 2026-07-24 · 10-Q · https://www.sec.gov/Archives/edgar/data/50863/000005086326000157/

[31] INTC 10-K filed 2026-01-23 · CCG market trends and competition · 2026-01-23 · 10-K · https://www.sec.gov/Archives/edgar/data/50863/000005086326000011/

[32] INTC 10-K filed 2026-01-23 · DCAI market trends and competition · 2026-01-23 · 10-K · https://www.sec.gov/Archives/edgar/data/50863/000005086326000011/

[33] INTC 10-K filed 2026-01-23 · foundry competition and capital intensity · 2026-01-23 · 10-K · https://www.sec.gov/Archives/edgar/data/50863/000005086326000011/

[34] INTC 10-K filed 2026-01-23 · Intel 14A external-customer condition · 2026-01-23 · 10-K · https://www.sec.gov/Archives/edgar/data/50863/000005086326000011/

[35] INTC 10-Q filed 2026-07-24 · Q2 2026 revenue drivers and supply constraints · 2026-07-24 · 10-Q · https://www.sec.gov/Archives/edgar/data/50863/000005086326000157/

[36] INTC 1Q26 earnings call 2026-04-23 · Q1 2026 segment performance and guidance · 2026-04-23 · earnings-call · https://www.intc.com/events-and-presentations

[37] INTC 10-K filed 2026-01-23 · FY2025 client and server revenue drivers · 2026-01-23 · 10-K · https://www.sec.gov/Archives/edgar/data/50863/000005086326000011/

[38] INTC 2Q26 earnings call 2026-07-23 · Q2 2026 foundry and product progress · 2026-07-23 · earnings-call · https://www.intc.com/events-and-presentations

[39] Yahoo Finance 2026-07-25 · Goldman delivers a candid response after Intel's stunning quarter · 2026-07-25 · Yahoo Finance · Mwangi Enos · https://finance.yahoo.com/markets/stocks/articles/goldman-delivers-candid-response-intels-193700306.html

[40] Wccftech 2026-09-18 · Intel's 14A process is being evaluated by Amazon, Apple, AMD, Google, Tesla, Microsoft, NVIDIA and Qualcomm, according to Piper Sandler · 2026-09-18 · Wccftech · Rohail Saleem · https://wccftech.com/intels-14a-process-is-currently-being-evaluated-by-amazon-apple-amd-google-tesla-microsoft-nvidia-and-qualcomm-according-to-piper-sandler/