traded at $103.61 on Tuesday, up $7.81 or 8.16%, with 55.087 million shares crossing by late morning against a three-month average of 113.398 million. That volume ranked it the single most actively traded stock in the United States on a session when the Dow Jones Industrial Average () shed 570.78 points and the S&P 500 () slipped 0.35% to 7,691.55.
The intraday range ran $96.04 to $103.24, with the print pushing marginally above the session high into the afternoon. Market capitalisation stands at $547.72 billion. The 52-week range spans $24.05 to $142.35, and the stock has appreciated 291.98% across twelve months.
Friday, September 4 set the base. Intel closed at $95.80 after climbing 4.5% on volume of 97.7 million shares. US markets were shut Monday for Labor Day, which concentrated three days of news flow into a single premarket window. The stock opened up 3.9%, topped the S&P 500’s premarket gainer list at +3.6%, and then more than doubled that move once cash trading began.
Three separate catalysts landed inside twenty-four hours. Supply-chain reporting indicated a roughly 10% increase to PC processor prices effective October 5. An upgrade to Outperform arrived with a $120 price target. And Intel Foundry disclosed that it has processed more than one million wafers using High NA EUV lithography.
None of those is a small item. Together they address the three questions that have defined the Intel debate for two years: can it price, can it manufacture, and will anyone else use its fabs.
The tape’s reaction was one-directional. The stock opened well below its eventual print and was accumulated steadily through the morning, closing the opening gap and extending. That kind of grinding, non-reversing advance in the market’s most liquid name is not short covering. It is institutional buying against a supply of stock that has been abundant since the company issued 210 million new shares last month.
Semiconductors broadly participated. Qualcomm (QCOM) rose 4.75% to $176.75. Oracle (ORCL) gained 3.39% to $164.16. Nvidia (NVDA), however, fell 1.46% to $227.00 — the market rotating within AI rather than into it.
The Price Hike: Roughly 10% on PC Processors, October 5
The primary catalyst is straightforward and it goes directly to gross margin. Supply-chain sources indicate Intel plans to raise PC CPU prices by approximately 10% on October 5, 2026.
That would be the third round of increases since the end of 2025. The first landed in the first quarter of 2026 at roughly 10%. The second came in July and covered selected consumer and server parts, with increases ranging from tens of dollars to more than a thousand.
The July round is worth examining because it shows the shape of the strategy. On the consumer side, the Core Ultra 7 270K Plus and Core Ultra 7 250K Plus rose between $30 and $50 depending on model, while the Core Ultra 9 285K held its $599 recommended price and some entry-level parts continued selling below launch. That is not a blanket cost pass-through. It is surgical pricing applied where demand is strongest.
The server side moved far more aggressively. Selected Xeon 6 Granite Rapids parts roughly doubled against mid-2025 retail levels. Certain Xeon 8000 Emerald Rapids models now carry recommended prices above their original launch references, with the largest increases exceeding $1,300. The flagship Xeon 6980P, a 128-core part, moved from $12,460 to $13,955 — a $1,495 increase, or roughly 12%.
Company statements attributed the adjustments to current market conditions, higher supply chain costs and demand exceeding supply for specific models.
The market read is what matters here. A company raising prices three times in twelve months without losing volume is a company with pricing power it did not previously possess. Intel spent the better part of a decade trading price for share against a lower-cost competitor. Pricing into shortage is the opposite behaviour, and it flows directly into the gross margin line that has been the single biggest bear argument on the stock.
Non-GAAP gross margin came in at 41.8% in the June quarter, 280 basis points ahead of guidance, with higher average selling prices from mix and pricing actions explicitly cited as a driver. Another 10% on PC parts compounds that.
Killing the Low-Margin Line Is the Bigger Signal
Buried alongside the pricing report is a strategic item that matters more over a multi-year horizon: chief executive Lip-Bu Tan is reviewing the low-margin Small Core product line, with some products potentially entering end-of-life.
Discontinuing product is a harder decision than raising price. It concedes revenue, market share and design-win presence in exchange for margin. Intel has historically been reluctant to do it, because share in the PC and entry-server market has been the foundation of the ecosystem argument — keep the installed base, keep the software, keep the moat.
Tan is trading that for profitability. Supply-chain sources describe the pricing actions and the product rationalisation as pointing at the same goal: raising overall gross margin and moving away from the strategy of trading price for volume.
The context makes it defensible. Market demand exceeded available product supply in the June quarter due to capacity constraints at Intel’s own factories and industry-wide shortages, and the company expects shortages of substrates, memory and other critical components to persist into next year. Full disclosure appears in the quarterly filing at sec.gov.
When supply is the binding constraint, every wafer allocated to a low-margin part is a wafer not allocated to a Xeon selling for $13,955. Rationalising the bottom of the portfolio is not a defensive retreat under those conditions. It is capacity reallocation toward the highest-return end of the mix.
The risk is well understood. Exiting the low end cedes ground to competitors who will take those sockets and may keep them when supply normalises. Intel is betting that the AI compute cycle runs long enough that the trade never has to be reversed.
Guidance already reflects the demand asymmetry. Management expects PC consumption to be sub-seasonal in the second half and down by low double digits for all of 2026 on memory prices and constraints, while the server CPU outlook has improved again, with strong double-digit industry unit growth forecast this year and next and momentum extending into 2028.
One Million Wafers on High NA EUV
The third catalyst is the manufacturing one, and it is the item that speaks to the foundry thesis rather than the product business.
Intel Foundry disclosed at an industry lithography conference that it has processed more than one million wafers using High NA EUV systems. Those machines cost up to $400 million each, and Intel has been the most aggressive early adopter of the technology in the industry.
Separately, the lithography supplier confirmed High NA EUV adoption commitments from Samsung, TSMC and Intel — the three companies capable of building leading-edge logic at scale. Shares of the equipment maker gained 2% on the announcement.
The one-million-wafer figure is the meaningful number. High NA has been dogged by scepticism about throughput, uptime and cost per wafer since the first tools shipped. A million wafers processed is production-scale validation rather than a demonstration, and Intel got there first.
That matters for the foundry pitch specifically. Intel’s argument to external customers is that it will have leading-edge capacity inside the United States with process technology at or ahead of the competition. High NA leadership is the most concrete evidence available that the second half of that claim is credible.
It also connects to 18A, the node built on gate-all-around transistors and backside power delivery. Intel has brought 18A to full scale with more than 400 Series 3 designs across consumer and commercial products, with high-volume production running at Fab 52 in Chandler, Arizona. Core Ultra Series 3, code-named Panther Lake, was the first product to ship on the node.
The chain of logic the market is pricing runs: High NA at scale, 18A in high-volume production, 14A on schedule, therefore a credible external foundry offering. Each link has now been demonstrated except the last one, which is the entire remaining question.
Capital spending on tools and clean-room space in the US from 2021 through 2026 is approaching $100 billion, substantially more than any other semiconductor manufacturer has committed domestically.
Technicals: $96.04 to $103.24 and the $142.35 Overhang
The chart has been violent in both directions, which is what a 291.98% twelve-month move looks like from the inside.
Tuesday’s action cleared a well-defined level. Intel had been building a staircase of higher lows and higher highs from a base near $85, closing around $95 to $96 across multiple sessions with several green days in the 3% to 5% range. The $95 level carries specific significance: it is the price at which 210,526,315 new shares were issued last month, meaning every buyer in that offering was at breakeven going into Tuesday.
Clearing $95 decisively, then $100, then $103.24, moves the entire offering book into profit. That removes a structural supply overhang — offering participants sitting at a loss are natural sellers into any rally, and they no longer are.
The overhead map from here is sparse. The 52-week high is $142.35, roughly 37% above Tuesday’s print, with no dense trading structure between the current level and the high $120s. The stock fell 28% during July and declined 11.22% over the subsequent month, which means the descent through this zone was rapid and left little accumulated volume behind.
Below, the levels are tighter. Tuesday’s low at $96.04 is immediate support and coincides with the $95 offering price. Friday’s $95.80 close sits inside the same band. A break below $95 would put the offering book back underwater and reopen the mid-$80s where the stock based before the current advance.
Volume is the caveat on the bullish read. At 55.087 million shares against a 113.398 million three-month average, Tuesday’s session ran well below normal turnover even at its peak activity. An 8% move on half-average volume is a thin move, and thin moves retrace more easily than heavy ones.
The 52-week low at $24.05 provides context for how far this has travelled. Anyone holding from that level is sitting on a 331% gain, and profit-taking pressure from that cohort has repeatedly capped rallies through 2026.
Relative strength has pushed toward overbought territory on the daily after an 8% single-session move, though it has not reached the extremes that historically preceded sharp reversals in this name.
Q2 by the Numbers: $16.1 Billion and the Fastest Growth Since 2011
The fundamental base underneath the current move is stronger than the stock’s reputation suggests.
June-quarter revenue reached $16.1 billion, up 25% year over year and $3.3 billion higher than the same period in 2025. That exceeded guidance by $1.8 billion at the midpoint and represented the fastest revenue growth rate for any quarter since 2011. It was the seventh consecutive quarter of beating financial expectations.
Non-GAAP earnings per share came in at $0.42 against guidance of $0.20 and consensus of roughly $0.21 — a beat of 100% against the company’s own forecast. Non-GAAP gross margin printed 41.8%, approximately 280 basis points ahead of guidance, driven by higher revenue, better factory yields and higher average selling prices from mix and pricing actions.
Year-to-date revenue reached $29.7 billion, up $4.2 billion from the first half of 2025. The company ended the quarter with roughly $30 billion in cash and short-term investments, before the August equity raise.
AI-driven businesses grew more than 70% year over year and contributed approximately 70% of total revenue. Purpose-built silicon revenue rose roughly 20% sequentially and nearly tripled year over year. Design services revenue grew nearly threefold.
September-quarter guidance calls for revenue of $15.8 billion to $16.8 billion, a $16.3 billion midpoint, with 42% gross margin, an 11% tax rate and $0.38 in non-GAAP EPS. Consensus entering the print was $15.1 billion and $0.27, so the guide arrived roughly 8% above on revenue and 41% above on earnings.
Non-GAAP operating expenses are being held to roughly $16.5 billion for the year. Non-controlling interest nets to approximately $250 million in each of the third and fourth quarters, rising to approximately $1.1 billion for 2027 and 2028 on a GAAP basis.
Full-year 2025 revenue was $52.9 billion, roughly flat, constrained by industry-wide supply. The 2026 trajectory — $29.7 billion through June plus a $16.3 billion guided third quarter — implies annual revenue in the low-to-mid $60 billion range, a step-change rather than a recovery.
Releases are posted at intc.com.
Data Center at $6.3 Billion and Up 59% Is the Real Engine
Strip the segments apart and one line explains almost all of the growth.
Data center and AI revenue rose 59% year over year to $6.3 billion in the June quarter, and 40% across the first half. Client computing revenue rose 13% year over year and 7% year to date. Intel Products revenue overall increased 28% from the prior-year quarter.
The increases came primarily from average selling price gains, the majority driven by a higher mix of premium products sold, with demand-based pricing actions contributing to a lesser extent to offset higher input costs.
That sentence is the whole margin story in company language. Intel is selling more expensive parts and charging more for them, and both effects are working simultaneously.
Xeon 6 has been one of the fastest-ramping products in company history, with year-over-year server growth the strongest on record. Management stated that data center operations cannot keep up with orders, leaving the company unable to fully meet customer demand.
The most durable detail is contractual. Intel has signed ten long-term agreements with server CPU buyers, structured around either fixed pricing commitments or guaranteed purchase volumes. Those contracts convert a cyclical business into something closer to a subscription, and they materially reduce the risk that the current pricing environment reverses when supply normalises.
The demand thesis rests on a specific architectural argument: as AI workloads shift from training toward inference and agentic systems, demand for general-purpose server CPUs rises rather than falls. Training is GPU-dominated. Inference and agentic orchestration require substantially more CPU per unit of accelerator, which is why the ratio of CPUs to GPUs in deployed systems is the number bulls point to.
If that shift is real, Intel’s data center franchise has a multi-year runway that has nothing to do with recapturing share from its direct x86 competitor. If it is not, the current growth rate is a supply-shortage artifact.
The company forecasts strong double-digit industry unit growth in server CPUs this year and next, with momentum extending into 2028.
The Foundry Problem: $5.8 Billion In, $293 Million External
The bull case has one hole, and it is large.
Intel Foundry generated $5.8 billion in June-quarter revenue, up 31% year over year. External foundry revenue within that figure totalled $293 million. The overwhelming majority of Intel Foundry’s revenue comes from manufacturing chips for Intel.
The segment posted a $2.1 billion operating loss in the quarter.
That is the arithmetic that has kept a large portion of the market sceptical regardless of how well the product business performs. A foundry serving primarily internal demand is a cost centre with a revenue line attached, and a $2.1 billion quarterly loss against $293 million of genuine external business is not a business — it is a bet.
The strategic history reflects how expensive that bet has been. Intel launched IDM 2.0 in March 2021, creating a foundry services arm and committing roughly $20 billion to two Arizona fabs while pursuing a five-node roadmap culminating in 18A. It relaunched the operation as Intel Foundry in February 2024 and won named 18A commitments from Microsoft and AWS.
The expansion has since been scaled back substantially. Planned fabs in Germany and Poland were cancelled. Ohio construction was slowed. Management initially indicated 14A could be paused without a major external customer before committing in 2026 to complete the node’s development.
The central unresolved question has not changed: whether Intel can secure a significant disclosed outside customer for 14A and convert foundry capacity into external business.
Early signals exist. SK Hynix is reportedly evaluating Intel Foundry for HBM4E base dies — advanced packaging and base-die work rather than leading-edge logic, but a real external workload from a memory leader. Design services revenue growing nearly threefold points in the same direction.
Capital expenditure guidance for 2026 was raised to more than $20 billion, with 2027 forecast significantly above that, the vast majority spent across the US network.
18A Is Shipping, 14A Is the Entire Thesis
The node roadmap is where the next three years get decided.
Intel 18A is in high-volume production at Fab 52 in Chandler, Arizona. It is the first Intel node with gate-all-around transistors and backside power delivery, and more than 400 Series 3 designs are in the market across consumer and commercial products. Panther Lake, marketed as Core Ultra Series 3, was the first 18A product, with the initial SKU shipping before the end of 2025 and broad availability from January 2026.
That is execution. Two years ago the market questioned whether Intel could deliver a leading-edge node at all.
Intel 14A is the node that determines whether the foundry becomes a business. Risk production is planned for 2027 with a high-volume ramp committed for 2028. It is the first node designed from the outset to serve external customers as well as internal products, and management has stated that 14A is ahead of where older technologies stood at the equivalent point in their cycles.
Landing a marquee external customer on 14A is the single most important catalyst available to this company. Reports have circulated that Nvidia is evaluating Intel’s foundry for its 2028 Feynman generation. If that materialises, it would validate the manufacturing roadmap more decisively than any other single data point, because Nvidia currently has no reason to use a second source except capacity and geography.
The $20 billion raised in August exists specifically to fund the gap between now and 2028. Capital expenditure exceeding $20 billion in 2026 and higher again in 2027, against a foundry losing $2.1 billion a quarter, requires a balance sheet that can absorb several more years of negative segment economics.
The pricing actions and the low-margin product rationalisation feed the same equation. Every basis point of gross margin improvement in the product business funds another quarter of foundry investment without additional dilution.
That is the structural link between Tuesday’s price-hike headline and the long-term thesis, and it is why an 8% move on a pricing report is not an overreaction.
Original Post