#英特尔Q2营收创15年最快增速 Intel Q2 earnings breakdown: Revenue up 25%, turning losses into profits—how to think about Intel after its comeback

First, one fact that nobody would have dared believe a year ago: over the past 12 months, Intel’s stock price has risen by more than 300%.

A year ago, this company was Wall Street’s most textbook “value trap”—falling behind on process nodes versus TSMC by two generations, its foundry business losing more than $10 billion in a single year, steadily getting eaten away in the server market by AMD, and being completely left in the dust by Nvidia in the AI wave. The market almost treated it as an old company waiting to be broken up, acquired, or slowly die. Then it came back to life.

Intel Q2 2026 earnings: Revenue was $16.13 billion, up 25% year over year, marking the strongest single-quarter growth rate in nearly 15 years; adjusted earnings per share were $0.42, well above expectations; Non-GAAP net profit was $2.2 billion, turning losses into profits. After the report, the stock price briefly surged more than 13% in after-hours trading.

A company that was in the ICU last year suddenly delivered a steep upward growth curve. What exactly happened in between? And more importantly: is this a genuine turnaround from real hardship, or a hype game propped up by a 120x trailing P/E? To answer that, you need to first understand three things: the business, the government, and valuation.

1. Three business lines moving up at the same time

Intel’s good quarter wasn’t propped up by just one line—it was all three lines moving higher simultaneously. Data Center and AI (DCAI) was the biggest surprise. Revenue hit $6.3 billion, up 59% year over year. This is the brightest number in the entire report.

In recent years, Intel’s server CPUs have been continually pushed to the ground by AMD’s EPYC, with market share steadily slipping. But the AI wave brought an unexpected tailwind: for every AI data center, besides filling it with Nvidia GPUs, you also need a lot of CPUs to handle host scheduling, data preprocessing, and general computing. As AI compute demand explodes, so does the CPU demand that supports it. Intel’s Xeon line is perfectly positioned to benefit from this “AI companion demand.”

Client Computing Group (CCG) stabilized the base business. Revenue was $8.9 billion, up 13% year over year. This is Intel’s traditional core business—processors for PCs and laptops. The AI PC upgrade cycle, combined with an overall rebound in the PC market, pushed this segment—once considered “no growth”—back into double-digit growth.

Intel Foundry (foundry services) is the narrative core. Revenue was $5.8 billion, up 31% year over year. This is the most critical—and hardest—piece of Intel’s entire comeback story. Gross margin improved to 41.8%. This number indicates that Intel isn’t only growing revenue—the quality of profitability is being repaired in a tangible way.

2. 18A: the make-or-break process node

Understanding Intel’s future is impossible without one codename: 18A. This is Intel’s 1.8-nanometer-class advanced process node—the ace it’s trying to catch up to TSMC with, and even surpass.

Over the past decade, Intel has fallen behind in process technology—7nm had delays, 10nm got pushed back—while TSMC turned the most advanced manufacturing into the world’s only choice for leading AI chips.

18A is Intel’s key battle in its “return to form.” And the latest progress is striking: 18A’s yield has climbed to about 85%. Yield is the lifeline of advanced process nodes. For comparison, TSMC’s most advanced 2nm (N2) yield is about 65%, and Samsung’s SF2 is about 40%. If Intel’s reported 85% yield data is real and can be ramped into stable mass production, it means that—on this process generation—Intel finally has the capital to go head-to-head with TSMC for the first time.

More importantly, demand-side customers are starting to place real orders. Intel management raised its 2026 capital expenditure plan from $18 billion to $20 billion, and expects 2027 spending to “significantly increase.” When a company dares to spend more, it usually means there are concrete customer commitments behind it. Market rumors and public information indicate that Google, Apple, and the Musk ecosystem (SpaceX, xAI, Tesla) are in talks with Intel for foundry services or have already signed.

If Intel can truly become a “second TSMC on U.S. soil,” the imagination space for this story is enormous. Every U.S. tech company worried about Taiwan Strait risks and seeking to de-risk its supply chain needs an advanced foundry facility outside of TSMC, located in the United States—and currently, the only company with a real shot at playing that role is Intel.

3. The most special shareholder: the U.S. government

In Intel’s story, there’s a variable no other chip company has: its second-largest shareholder is the U.S. government.

In August 2025, the Trump administration announced it would convert the subsidies Intel received under the Biden-era “CHIPS Act” into an equity investment. At a price of $20.47 per share, the U.S. government invested $8.9 billion to buy 433.3 million shares, taking about 10% of Intel.

This is an extremely rare event in U.S. commercial history. The federal government directly became a major shareholder of a public technology company. Even though this is “passive ownership” (no board seats, no governance interference), its symbolic significance is huge.

It means two things.

First, Intel now has “national team” backing. When the U.S. government sits on a company’s shareholder register, the ability to win government orders, defense chip orders, and to capture industrial policy tilt is something no competitor can match. If the U.S. wants to rebuild domestic semiconductor manufacturing capacity, Intel is the selected “chosen child.” This political capital is Intel’s hardest ace in its turnaround-from-difficulty narrative.

Second, it also brings controversy and risk. Government ownership triggers intense debate about “national intervention in private enterprises,” and some lawmakers have publicly questioned the deal. As a shareholder, will the government influence Intel’s business decisions in the future? Could Intel be forced into uneconomical investments due to political factors? These questions remain unresolved.

In addition, Nvidia also made a strategic investment of $5 billion to buy into Intel—an intriguing signal. Former rivals have now become investors. Nvidia needs a production capacity backup beyond TSMC; Intel needs Nvidia’s orders and backing. The two parties quickly aligned. With one company held by both the U.S. government and Nvidia, Intel has unique political and industry support in the global semiconductor sector.

4. Competitive positioning: where does Intel actually stand?

To assess Intel clearly, you must place it back into the real competitive landscape. On AI GPUs, Intel is still absent. This is its biggest weakness.

The core of AI compute is GPUs, and this market is dominated overwhelmingly by Nvidia (70%–80%). AMD is the second. Intel’s Gaudi AI accelerators basically have no meaningful presence. On the most core and most profitable “cake” of AI, Intel currently can’t get a seat at the table. What Intel is benefiting from is “AI companion demand” (CPU), not “AI core demand” (GPU).

This positioning difference sets its ceiling.

In server CPUs, Intel is on the defensive-counterattack path. Versus AMD’s EPYC, Intel’s Xeon finally halted share losses over the past two years, driven by the overall expansion of AI data centers, delivering 59% growth. But AMD remains a strong opponent—this is a hard, head-to-head fight.

In foundry services, Intel is chasing TSMC. This is Intel’s heaviest bet and also the most imaginative direction. TSMC remains the absolute king (over 72% share in global leading-edge process nodes). But with Intel’s 18A yield breakthrough, plus the triple narrative of “U.S. domestic + government backing + supply chain de-risking,” Intel has, for the first time, a chance to take a small piece of meat from TSMC’s mouth.

Note: a “small piece.” In the near term, Intel can’t shake TSMC’s position. But a 0-to-1 breakthrough in itself is enough to support a period of valuation expansion.

In PCs, Intel is a defender. The AI PC upgrade cycle gave it some breathing room, but the PC market is mature—providing stable cash flow without delivering high growth. In one sentence, Intel’s ecosystem position is this: it is missing from the AI core battleground (GPUs), but it has found its place in the AI companion battleground (CPUs) and the infrastructure battleground (domestic foundry services). And with government backing, it has gained political capital that others can’t.

5. Valuation: 120x PE—are you buying fantasy or the future?

Now for the most painful part: valuation.

Over the past year Intel’s stock has surged more than 300%, and its market cap has topped $600 billion. The current P/E ratio is over 120x.

What does a 120x PE mean? For a company like TSMC, with a 45% net profit margin, global monopoly in leading-edge process nodes, and CoWoS packaging, its PE is only around 30x. Intel using a 120x valuation indicates that the market isn’t really buying what Intel is earning today—it’s pricing in a full set of fantasies for the next three to five years: the foundry becoming broadly profitable, taking back share from TSMC, and becoming a national pillar of U.S. semiconductor manufacturing.

This valuation does have a logic. The hallmark of “turnaround stocks” is that when the market believes “the worst is already behind us and the turning point has arrived,” it front-loads multi-year good expectations all at once into the stock price.

Intel’s 300%+ surge this year is the extreme expression of this “expectation re-pricing.” But a 120x PE also implies extreme fragility. It has priced in too many good things upfront. If any link breaks—18A mass production underperforms, big-customer orders fail to materialize, the foundry once again falls back into losses, or AMD launches a counterattack in servers—valuation could unwind sharply. There’s no margin for error at this price.

The Q2 earnings report jumped 13% after hours, showing the market is still willing to keep buying into this story—for now.

Third-quarter guidance is also strong (revenue $15.8–$16.8 billion, adjusted EPS $0.38), both above expectations. Near-term momentum is upward. But momentum and valuation are two different things. Momentum can push the stock price higher, while valuation determines how bad the drop can be once the narrative cracks.

6. My view on Intel

Start with the conclusion: this is a real turnaround from genuine hardship, but the current stock price has already priced in most of the turnaround benefits.

As for the authenticity of the turnaround—I lean toward believing it.

Three business lines moving up together, gross margin repair, turning losses into profits, 18A yield breaking through 85%, Intel’s double entry by both the government and Nvidia, and big-customer orders landing one after another—these are not financial window dressing. They reflect real fundamental improvements.

Intel’s worst days are very likely behind it. After CEO Lip-Bu Tan took over, a series of focused actions on process nodes and foundry services are starting to pay off.

On valuation—stay cautious. A 120x PE and a 300% gain in one year mean this stock has already moved from a “severely undervalued value stock” to a “fully priced—and even a bit overheated—growth story.” At this level, you’re no longer buying a “cheap good company,” you’re buying an “expensive good story.” Whether the story can be delivered requires step-by-step verification over the next two or three years—and each step has the risk of failure.

If you don’t already hold it, this isn’t a comfortable entry point. Chasing a 120x PE turnaround stock is an action with a poor risk-reward profile.

A more rational approach is to wait for a pullback triggered by some short-term negative catalyst (for example, a quarter’s guidance missing expectations, or a foundry customer slipping away). After market sentiment cools and some of the valuation gets digested, then enter in batches. Don’t rush in when it has already risen 300% in a year and everyone is discussing it.

If you already hold the stock and are sitting on solid profits, you could consider taking some profits to bring down your cost basis, then continue holding the remaining position to participate in this turnaround story. That locks in realized gains while not fully missing the future upside implied by the narrative.

The three variables that matter most going forward:
First, whether 18A’s mass production ramp and yield can be stabilized at high levels—this is the technical prerequisite for the foundry story to stand.

Second, whether foundry big-customer orders can move from “signed” to “ramped volume.” Names like Google, Apple, and Nvidia ultimately need to show up in real revenue numbers to count.

Third, whether the foundry business can reach breakeven around 2027—this is the financial endpoint that turns the turnaround narrative from “imagination” into “reality.”

Intel’s revival is real. But the market has paid a high price for this revival already. It’s a good company now, but it may not be a good price. The most profitable phase of a turnaround is buying when nobody cares and the market is desperate—that phase, a year ago, was already behind us. People entering now aren’t betting on “will it survive,” but on “can it deliver the larger future implied by a 120x PE.” Those are two completely different levels of difficulty. #夏日创作营