INTC: Best Quarter in 15 Years, Down 37%

Western semiconductor manufacturing is approaching a genuine inflection point, and the company at the center of it just posted its fastest revenue growth in nearly fifteen years, watched its stock fall 37% from peak, and still has not named the foundry customer that would resolve the entire debate.

That is the Intel situation entering August 2026. The CPU business is recovering faster than most analysts projected twelve months ago. The foundry is losing $2.1 billion a quarter, improving sequentially, and carrying a customer pipeline that has been described, hinted at, and partially confirmed, but not yet disclosed at the level institutions need before they will reprice the stock. The distance between those two realities is where the investment thesis lives.

The question this report is built to answer is not whether Intel’s CPU business has recovered. That is settled. The question is whether the foundry deserves a valuation, and if so, when the market will begin assigning one. The answer points to October 21, 2026.

The Market Opportunity

The global semiconductor industry is undergoing a structural realignment with no modern precedent. AI infrastructure spending is pulling capital into chip manufacturing at a scale that even the largest foundries are struggling to absorb. Omdia projected a 94.1% year-over-year surge in global semiconductor revenue for 2026. That is not an early-year forecast. It is the trajectory the industry is already on.

Within that environment, two things are happening simultaneously. AI server CPU demand is running ahead of supply. Intel said on its Q2 call that data center operations cannot keep pace with orders, constrained by substrate and memory shortages rather than weak demand. That is a structurally different problem from every prior Intel cycle, when excess inventory was the chronic operational concern.

The second development is more important for the long-term thesis: U.S. and European governments are actively subsidizing the creation of a domestic semiconductor manufacturing alternative to TSMC. Intel is the only Western company positioned to fulfill that role at leading-edge scale. The U.S. government holds approximately a 10% stake in Intel, acquired through the CHIPS Act Secure Enclave Agreement at roughly $20 per share. That policy backstop has no equivalent among Intel’s peers. The structural logic behind it, ensuring that critical semiconductor manufacturing does not depend entirely on a single geography, has not weakened.

The opportunity for investors is specific: the market is currently pricing Intel as a CPU recovery story with a foundry business worth approximately zero. If the foundry closes even one high-volume, publicly named external customer on its 18A node, that valuation changes. The entire debate compresses into a single catalyst window.

Why This Company

Intel’s Q2 2026 results were exceptional by any measure applied to the income statement. Total revenue of $16.1 billion, up 25% year-over-year, beat the LSEG consensus of $14.42 billion by $1.68 billion. Non-GAAP EPS of $0.42 doubled the $0.21 estimate. The stock fell 7.89% the following session and has since extended the decline to more than 37% from its June 22 all-time closing high of $140.94.

When a stock falls on a beat of that magnitude, the market is not rejecting the income statement. It is pricing a future the income statement does not yet contain. For Intel, that future is external foundry revenue at volume on 18A.

The Data Center and AI segment delivered $6.3 billion in Q2 revenue, up 59% year-over-year. Operating income in that segment grew from $600 million in Q2 2025 to $2.5 billion in Q2 2026. Operating margin expanded from 16.1% to 39.5% in twelve months. Intel described Q2 year-over-year server growth as the strongest on record for the company, with Xeon 6 characterized as one of the fastest-ramping products in its history. Intel also signed ten long-term contracts with server CPU buyers in Q2, structured around fixed pricing or guaranteed purchase volumes. A supply security development that rarely earns a headline, but it has meaningful cash flow implications over the next two to four years.

The Client Computing Group posted $8.9 billion in Q2 revenue, up 13% year-over-year, driven by higher PC volumes and improved average selling prices. Management flagged that flat PC sales in Q3, due to a memory shortage, could weigh on that segment in the current quarter — the risk embedded in the $15.8 to $16.8 billion Q3 guidance range. Good guidance, but with a ceiling partly constrained by supply.

Intel Foundry is where valuation tension concentrates. Segment revenue of $5.8 billion, up 31%, sounds significant. The detail that matters: only approximately $293 to $300 million of that total came from external customers. The remaining $5.5 billion is intersegment revenue from Intel Products, eliminated in the consolidated figure. The foundry operating loss was $2.1 billion. That same segment lost $3.2 billion at a negative 71.7% operating margin in Q2 2025. The direction of travel has improved by more than $1 billion year-over-year while node execution has advanced. The market has not yet agreed to price that trajectory.

On process technology: Intel Foundry entered high-volume manufacturing for a subset of Core Ultra Series 3 processors using ASML’s High-NA EUV lithography. Intel is the first company delivering high-volume logic chips using that tool. ASML confirmed this on its July 15 earnings call. 18A-P entered risk production during Q2, meeting the timeline Intel had shared with customers and partners the prior year. The 18A yield rate improved to approximately 85%, up from 65% the prior quarter, with output running more than 50% above the prior quarter’s level.

Competitive Advantage

Intel’s competitive position is unusual because it operates in two markets simultaneously, and those markets reward different qualities. In CPUs, Intel competes on performance, power efficiency, and ecosystem compatibility. In foundry services, it competes on process technology, yield, cost per wafer, and customer relationships. Most companies do not have to win both arguments at the same time. Intel does.

The CPU moat is recovering. Xeon 6’s ramp rate reflects a product cycle that is working. AI infrastructure buildout is pulling demand toward x86 server architecture in a way that benefits both Intel and AMD, and Intel’s supply constraint is a demand validation rather than an operational failure. The ten long-term supply agreements signed in Q2 extend revenue visibility well beyond the current quarter.

The foundry customer evidence is layered. Microsoft confirmed an 18A order for its Maia AI accelerator program. Amazon signed a deal to use 18A for a new AI fabric chip. Fortinet became the first publicly named external foundry customer for Intel 4. The Apple angle carries additional weight: President Trump stated in June 2026 that Apple agreed to work with Intel to design and manufacture chips in the United States. Neither Apple nor Intel confirmed a formal foundry contract. A presidential statement is not a signed agreement. But it placed the question on every institutional desk in a way a typical analyst report cannot.

The financial position is stronger than the GAAP figures suggest. Intel recorded an $11 billion GAAP net loss in Q2, driven almost entirely by a $12.5 billion non-cash charge for mark-to-market valuation on escrowed CHIPS Act shares. Adjusting for that item, non-GAAP net income was approximately $2.2 billion. Non-GAAP gross margin came in at 41.8% and non-GAAP operating margin at 17.2%. The company is not in operational distress. Its accounting complexity is obscuring a business generating real operating income at scale.

Capital allocation is the sharpest point of concern. Intel raised its 2026 capital expenditure guidance from $18 billion to more than $20 billion, with 2027 spending described as a further meaningful increase. The CPU business is generating strong operating cash flow, but not yet enough to fund that buildout organically. The government stake and CHIPS Act funding provide a structural backstop, but the market reads rising CapEx without proportional external foundry revenue as a cash burn risk that has not been resolved.

Risks

The primary risk is straightforward: if October 21 arrives and Intel still cannot name a significant 18A external customer at volume, the foundry’s valuation contribution remains near zero, and the current stock price still embeds optionality that would need to unwind. A $2.1 billion quarterly operating loss in the foundry segment, sustained without meaningful external revenue growth, becomes progressively harder for institutional holders to justify against a capital allocation discipline benchmark.

The TSMC competitive risk arrived with new specificity during the week of July 31. The Information reported that TSMC is developing advanced chip packaging technology described internally as similar to Intel’s Embedded Multi-die Interconnect Bridge technique. If confirmed as a direct competitive product, it narrows one of Intel Foundry’s most technically defensible advantages. TSMC entering that market would not eliminate Intel’s packaging capabilities, but it would reduce the premium those capabilities command in customer conversations.

The 14A schedule risk is longer-dated but consequential. Intel reaffirmed 14A high-volume manufacturing by 2028, with CFO David Zinsner stating the process is ahead of where older technologies were at the same point in their development cycle. Any slip on that schedule extends the window of TSMC dominance in leading-edge manufacturing. Management previously flagged that 14A could be paused if external customer demand falls short of projections. That conditional creates a second-order risk: if 18A customer traction disappoints, 14A investment discipline comes into question simultaneously.

The macro calendar adds two near-term inputs. The July nonfarm payroll report lands August 8, with consensus at approximately 87,500 jobs added, up from 57,000 in June. A significant miss could be read as softening AI infrastructure spending expectations, weighing on capital-intensive names like Intel. AMD reports August 5, and that result will either confirm or challenge the durability of the AI server CPU demand cycle driving Intel’s 59% DCAI growth. Both events land before the next meaningful Intel-specific catalyst.

What to Watch

Four developments will determine whether the foundry thesis advances or stalls between now and October 21.

First: a named, high-volume 18A external customer. This is the single most powerful catalyst available to the stock in the next 60 to 90 days. Microsoft and Amazon have engaged Intel on 18A for specific programs. Apple’s involvement remains publicly unconfirmed as a formal foundry contract. Any formal disclosure with volume commitments, from any of those names or a new one, would directly challenge the market’s current assumption that the foundry is worth zero.

Second: the Q3 earnings call on October 21. Intel guided Q3 revenue to a $16.3 billion midpoint, beating the prior Wall Street consensus of $15.1 billion and marking the seventh consecutive quarter in which Intel’s own guidance proved conservative. The market’s question for that call is not the revenue line. It is whether foundry external revenue shows meaningful sequential growth from the $293 million Q2 baseline, and whether a significant 18A customer can be named publicly. Those two items are the gating factors for a sustained re-rating.

Third: AMD’s August 5 result. Data center CPU share is a real contest between Intel and AMD. If AMD’s data center segment shows strong year-over-year acceleration, it raises questions about Intel’s 59% DCAI growth as a sustainable run rate. If AMD decelerates, Intel’s numbers look more structurally durable. Either way, the read matters for how institutions size Intel exposure into the Q3 call.

Fourth: any update on the TSMC packaging development. The Information reported it on July 31. No formal TSMC announcement has followed. If TSMC confirms a commercial product that directly competes with Intel’s EMIB, institutional reassessment of Intel Foundry’s packaging differentiation should be expected quickly. If the report proves inaccurate or overstated, that overhang clears and removes one of the more recent arguments against the foundry thesis.

Final Verdict

Intel’s position entering August 2026 is one of the more genuinely asymmetric situations in large-cap technology. The CPU business is operating at its highest revenue growth rate since 2011, with margin expansion outpacing analyst estimates for seven consecutive quarters. The foundry is losing $2.1 billion per quarter, improving at a pace of more than $1 billion year-over-year, and holding a customer pipeline that includes Microsoft, Amazon, and an unconfirmed but repeatedly referenced Apple engagement. The stock is 37% below its all-time high despite a quarter that beat revenue consensus by $1.7 billion.

The 48-analyst consensus rates INTC a Hold with a 12-month average price target of $115.27, implying approximately 28% upside from the current $89 to $90 range. Wedbush raised its target to $98 from $60 post-earnings. A buy rating was reaffirmed at a $160 target as recently as July 29. The range of analyst opinion is wide, which reflects the binary nature of the foundry outcome more than any disagreement about the CPU business.

The investment case rests on one proposition: the foundry’s commercial trajectory, improving yields, rising output, a confirmed cloud provider commitment on 18A, and a Q3 earnings call that could carry a named customer announcement, is underpriced relative to the capital Intel has committed and the government policy surrounding it. A named 18A customer at volume, disclosed on or before October 21, is the single event most capable of closing the gap between Intel’s operational trajectory and the price at which the stock currently trades.

The risk is equally defined. If October arrives without that disclosure, the foundry optionality embedded in the current price unwinds, and the stock’s multiple compresses further. Investors with a multi-quarter horizon who can hold through the October catalyst window are buying a business generating $2.5 billion in quarterly operating income from its data center segment alone, at a price that assigns the foundry a value of approximately zero. That gap between what the CPU earns and what the market prices is the thesis. Whether it resolves in one quarter or three is the only open question.