Three chipmakers that spent years playing supporting roles to Nvidia just had the best quarter of their public lives. Micron, Intel, and AMD added a combined $2 trillion in market capitalization during the second quarter of 2026, according to CNBC’s analysis of trading data through late June. Micron’s stock climbed 240%. Intel gained 216%. AMD came close to tripling, up 186%. The rally hasn’t cooled since: by mid-August 2026, AI chip demand had helped push the S&P 500 to an intraday record of 7,814.88, with semiconductor names cited as the key driver, according to Intellectia AI.
The chip stock rally marks a real shift in how investors are pricing artificial intelligence hardware. For most of 2024 and 2025, Nvidia absorbed nearly all of the enthusiasm around AI chips. By the second quarter of 2026, that money started spreading out: into memory makers racing to supply high-bandwidth memory, into Intel’s long foundry turnaround, and into AMD’s push for a bigger slice of the data center. The VanEck Semiconductor ETF, which tracks the broader sector, rose 196% in the quarter, its strongest three-month stretch since it went public, not 71%. The rotation didn’t stop at quarter’s end: the Philadelphia Semiconductor Index climbed another 18% from its July 29 low through mid-August 2026, according to the Wall Street Journal, and was up roughly 10% for the month alone — its best August performance since 2003, per PrimeXBT. The SOXX ETF’s year-to-date return reached as high as 70-83% by mid-August, PrimeXBT reported, underscoring how durable the rotation into memory, foundry, and data-center chipmakers has been.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
Chip Stocks Add $2 Trillion in a Single Quarter
The numbers behind this chip stock rally are large enough to reshuffle how Wall Street ranks its biggest technology companies. Micron added roughly $920 billion in market value. Intel added about $480 billion. AMD added close to $615 billion. Combined, the three companies moved into the upper ranks of the most valuable U.S. tech firms by market capitalization, per CNBC’s reporting on the quarter.
None of this happened because of one announcement. It built over months, through earnings calls, foundry updates, and a steady drumbeat of AI data center spending from Microsoft, Google, Amazon, and Meta. What changed in the second quarter is that investors stopped treating Nvidia as the only way to bet on AI hardware. The hardware sector now includes memory suppliers and foundries as legitimate, stand-alone trades rather than Nvidia’s supporting cast.
Micron led the pack by a wide margin, and the setup for that move was visible months earlier. Micron’s fiscal second-quarter earnings, reported March 18 and covered here at the time, already showed record revenue of $23.9 billion. That report, built on memory prices that kept climbing as data centers scrambled for capacity, set the tone for a stock that would go on to gain 240% by the time the calendar quarter closed.
Micron’s 240% Surge: Inside the HBM Sellout
Micron’s rally traces back to a supply-demand imbalance that has been building for two years. High-bandwidth memory, the specialized DRAM that sits next to AI accelerators and feeds them data, is sold out. Micron said its entire 2026 HBM output was already committed under multi-year contracts, a point repeated across multiple analyst notes in the first half of the year.
That scarcity handed Micron pricing power it hadn’t had in over a decade. Its fiscal second-quarter earnings, reported by CNBC on March 18, showed revenue climbing to $23.9 billion. Micron’s own presentation materials described AI as recasting memory into what the company called a defining strategic asset for the industry, not just another commodity input. S&P Global Market Intelligence’s analysis of the results pointed to DRAM as the single biggest source of upside, with HBM as the driver underneath it.
A $200 Billion Bet on Capacity
Micron isn’t just riding the wave. The company committed roughly $200 billion to new manufacturing capacity, a bet aimed at locking in HBM and DRAM supply for years to come. Gross margins reportedly expanded to nearly 74.9%, up from about 39% a year earlier, according to one analysis of the quarter’s results, which is as clean a sign of pricing power as a memory maker gets. Micron lived through the opposite side of this cycle in 2022 and 2023, when memory prices collapsed. This time, the shortage is working in its favor.
Intel’s Comeback: 18A Progress Meets a Government Stake
Intel’s rally has a different shape. Where Micron’s story is about pricing power, Intel’s is about proving a five-year turnaround is finally working. Its 18A manufacturing process, the company’s most advanced node, moved from promise toward production through the first half of 2026. Panther Lake, marketed as Core Ultra Series 3, launched at CES in January and began shipping on 18A silicon.
Intel also shifted how it talks about its foundry business. In March, CFO David Zinsner said CEO Lip-Bu Tan was growing more open to offering 18A capacity to outside customers, a reversal from Intel’s earlier internal-only approach, according to Reuters. At Computex in June, Intel unveiled the Crescent Island AI inference GPU, its clearest signal yet that it wants a piece of the AI accelerator market Nvidia and AMD have dominated, with support for up to 480GB of LPDDR5X memory, according to Tom’s Hardware’s live coverage of the show. The same event produced Intel’s Xeon 6+ “Clearwater Forest” data-center chips, which the company says will pack up to 288 cores.
Then there’s the government stake. The U.S. took a 10% equity position in Intel in 2025 as part of a push to keep advanced chip manufacturing onshore, a move that gave the stock a floor even before the AI-driven rally arrived. That stake, combined with an earlier stretch of gains in April tied to Intel’s foundry ambitions, had already put the stock on a recovery path before the second quarter accelerated it into triple-digit territory.
AMD Rides Record Server Share to a 186% Gain
AMD’s gain looks smaller next to Micron and Intel, but a 186% quarterly move is still remarkable for a company AMD’s size. Much of the story centers on the data center. AMD’s Epyc server processors kept taking share from Intel’s Xeon lineup, and by one industry tracker’s account, AMD held roughly a third of the server CPU market by the second quarter, a record for the company.
AMD also used the quarter to lock in its desktop platform’s future, confirming that its AM5 socket would remain supported through 2029. That kind of commitment matters more to investors than it sounds, since platform longevity keeps enterprise and consumer customers from jumping to a competing architecture. On the infrastructure side, AMD reportedly committed roughly $2.67 billion toward AI research and data center buildout in the United Kingdom in June, part of a broader push to land AI infrastructure deals outside the U.S.
None of it matches the scale of AMD’s Instinct GPU business, now the company’s most closely watched growth line. AMD’s most recent earnings before the rally showed its data center segment pulling in $5.8 billion, and investors spent the second quarter betting that figure keeps climbing as the MI-series roadmap matures. That bet paid off in July 2026, when AMD used its Advancing AI 2026 event to launch the Instinct MI455X GPUs alongside new Helios rack-scale systems, according to HPCwire, giving the MI-series roadmap its clearest hardware milestone yet.
Q2 2026 Chip Stock Rally at a Glance
The table below lines up how the sector performed during the quarter, using CNBC’s reporting as the baseline for the three largest movers and additional detail from a Moneycheck analysis of the broader rally.
| Company | Q2 2026 Stock Gain | Market Cap Added | Primary Driver |
|---|---|---|---|
| Micron | +240% | ~$920 billion | HBM sellout, DRAM pricing power |
| Intel | +216% | ~$480 billion | 18A foundry progress, government stake |
| AMD | +186% | ~$615 billion | Record server CPU share, Instinct GPU demand |
| Marvell Technology | +200% | Not disclosed | Custom AI silicon for hyperscalers |
| Arm Holdings | +134% | Not disclosed | Licensing growth tied to AI chip design |
| Nvidia | +15% | Not disclosed | Already priced for AI dominance |
| VanEck Semiconductor ETF (SMH) | +71% | N/A | Broad sector rotation, best quarter since 2000 |
Source: CNBC (June 30, 2026), plus a Moneycheck analysis of Q2 2026 trading data.
Why Nvidia Sat Out the Rotation Trade
Nvidia’s absence from the top of the leaderboard doesn’t mean it had a bad quarter. The company remains the largest and most profitable AI chip supplier by a wide margin, and it posted an $81.6 billion quarter with data center revenue up 92% earlier in 2026. Nvidia even used the period to push into new territory, unveiling the RTX Spark platform with Microsoft and MediaTek at Computex on June 2, a chip built for AI-capable Windows PCs that packs 1 petaflop of AI performance and up to 128GB of unified memory, according to Nvidia’s own announcement.
What Nvidia didn’t get was a re-rating. Its stock rose about 15% in the quarter, a healthy gain by most standards but a rounding error next to Micron’s 240%. The simplest explanation is that Nvidia was already priced as the default AI infrastructure bet. Micron, Intel, and AMD had more room to run because the market hadn’t fully priced in their roles in the same buildout. That gap is exactly what this chip stock rally closed.
Beyond the Big Three: Marvell, Arm, and the Rest of the Sector
Micron, Intel, and AMD dominate the headlines, but they weren’t the only chip stocks moving in the second quarter. Marvell Technology, which builds custom AI silicon for hyperscale customers, gained roughly 200% in the same period. Arm Holdings, whose chip designs sit underneath most of the industry’s custom silicon efforts, rose about 134%. The foundry business behind nearly all of this hardware told the same story: TSMC, which manufactures chips for Nvidia, AMD, and most of the industry, posted Q2 2026 revenue of $40.2 billion, up 36% year-over-year, with AI and high-performance computing accounting for 66% of wafer sales, per TechTimes — a continuation of a run that saw first-quarter profit jump 58% to $18.2 billion and its shares climb 35% year-to-date by April 2026, according to Reuters.
The common thread is exposure to AI infrastructure that isn’t Nvidia’s own GPU lineup. Marvell’s custom silicon business serves cloud providers building their own accelerators, a category that includes projects like Google’s TPU program. Arm’s licensing model earns a small fee on nearly every chip design in the AI stack, from data-center CPUs to the phones running on-device AI features, which makes it a lower-risk way to bet on the same trend. Broadcom occupies the same lane: its AI chip revenue hit $10.8 billion in the June 2026 quarter, up 143% year-over-year, helping push total revenue to $22.19 billion, according to Yahoo Finance. Even China’s SMIC, hemmed in by export controls, posted Q2 2026 revenue of $3.01 billion against a $2.85 billion forecast and earnings per share of $0.06 versus expectations of $0.0372, Investing.com reported in August 2026. Investors spreading money across all of these names during the quarter suggests the AI infrastructure trade has broadened well past a two-company race between Nvidia and everyone else.
The HBM4 Bottleneck Reshaping the AI Supply Chain
High-bandwidth memory sits at the center of almost every storyline in this rally. AI accelerators need it to move data fast enough to keep expensive GPU and CPU silicon from sitting idle, and there currently isn’t enough of it to go around. Nvidia reportedly certified Samsung, SK Hynix, and Micron as qualified HBM4 suppliers in early June, a milestone that effectively determines who gets to sell memory into Nvidia’s next generation of accelerators.
Samsung and SK Hynix Race to Catch Up
Samsung’s HBM4 revenue reportedly crossed $1 billion for the first time by late June, a sign the company is closing the gap it opened when it fell behind Micron and SK Hynix earlier in the HBM cycle. OpenAI’s own custom silicon plans, which reportedly lean on a Samsung HBM4 supply agreement, show how far the memory shortage has spread beyond the traditional GPU makers and into the AI labs themselves.
The scarcity has consequences beyond stock prices. Data center operators building capacity for Microsoft, Google, Amazon, and Meta are competing for the same limited pool of HBM, and Micron’s own materials describe AI and server memory demand together crossing half of the entire memory industry’s addressable market for the first time in 2026. That kind of demand doesn’t resolve itself in a single quarter, which is part of why Micron, Samsung, and SK Hynix are all pouring capital into new capacity rather than waiting out the cycle.
What the Rally Means for AI Infrastructure Spending
For companies building AI data centers, this chip stock rally confirms a cost problem they already knew about. Memory and compute together make up the bulk of AI infrastructure spending, and if HBM stays sold out into 2027, hyperscalers will keep paying premium prices or scale back build plans. Data center lease commitments from Meta and Microsoft alone reportedly topped $850 billion earlier in 2026, a figure that assumes chip and memory supply keeps pace with construction schedules.
For the chip industry itself, the rally signals that Wall Street now treats AI infrastructure as a multi-vendor market instead of a Nvidia monopoly. That should, in theory, encourage more competition and eventually more supply. The timelines involved in building new memory fabs or ramping foundry capacity run into years, not quarters, though. Micron’s $200 billion capacity commitment will not fully show up in shipped product until well into the back half of the decade, which means today’s scarcity pricing has a long runway before new supply can meaningfully cool it off.
How This Compares to Past Semiconductor Cycles
Chip investors have watched booms turn into busts before, and the pattern is worth remembering here. Global semiconductor sales hit a record $555.9 billion in 2021, up 26.2% from the prior year, according to Semiconductor Industry Association data. That boom, driven by pandemic-era shortages, carried into 2022, when sales rose again to $574.1 billion, though growth had already slowed sharply in the second half of that year.
Then came the bust. Semiconductor sales fell 8.2% in 2023 to $526.8 billion, as memory prices collapsed and data center customers worked through inventory they had over-ordered during the shortage years. The industry recovered in 2024, with sales climbing 19.1% to $627.6 billion according to SIA’s market data, setting up the conditions that fed into 2026’s AI-driven rally.
| Year | Global Semiconductor Sales | YoY Change | Cycle Phase |
|---|---|---|---|
| 2021 | $555.9 billion | +26.2% | Pandemic shortage boom |
| 2022 | $574.1 billion | +3.3% | Record year, slowing in H2 |
| 2023 | $526.8 billion | -8.2% | Memory downturn |
| 2024 | $627.6 billion | +19.1% | AI-driven recovery |
Source: Semiconductor Industry Association (SIA).
The lesson from that cycle is that memory pricing power can vanish as fast as it appears. Micron lived through the 2022-2023 downturn directly, watching HBM and DRAM prices crater before AI-driven demand arrived to absorb the oversupply. Nothing about the 2026 chip stock rally guarantees a different ending, only a different cause behind the current shortage.
Micron vs Intel vs AMD vs Nvidia: Who Wins From Here
Each of the four companies is now playing a different game. Micron has the cleanest story: it sells a component in severe shortage, and pricing power should persist as long as HBM stays sold out. The risk is cyclicality. Memory has crashed before, and it can crash again once supply finally catches up with demand.
Intel has the longest road. Its rally depends on 18A yields staying strong and, eventually, on outside customers actually paying for foundry capacity rather than just discussing it. External foundry revenue remains a small fraction of Intel’s total business, so the turnaround story still needs proof rather than momentum alone.
AMD sits in between. Its server CPU share gains are real and measurable, but its AI GPU business still trails Nvidia by a wide margin in scale, even as it grows quickly off a smaller base. The MI-series roadmap gives AMD a credible second-place position in AI accelerators, which is a genuine business but a different one from Nvidia’s.
Nvidia remains the company every one of these three gets measured against. It didn’t post the same percentage gain, but it starts from a base so large that a modest percentage move still represents tens of billions of dollars. Betting against Nvidia’s role in AI infrastructure has been a losing trade for three straight years, and nothing in this quarter’s numbers changes that.
The Bubble Question: What Skeptics Are Watching
Not everyone is comfortable watching stocks triple in three months. One analysis of the rally noted that individual stock volatility in the sector reached levels not seen since 2015, a sign that some of the move is driven by momentum trading rather than fundamentals catching up to price. When three companies add a combined $2 trillion in a single quarter, valuation discipline tends to take a back seat to fear of missing out.
The bear case rests on a few specific risks. HBM shortages could ease faster than expected if Samsung and SK Hynix ramp capacity ahead of schedule, cutting into Micron’s pricing power. Intel’s foundry business could keep losing money on external customers longer than the market is currently pricing in. AMD’s data-center GPU business, while growing, still needs to prove it can take meaningful share from Nvidia rather than simply expanding alongside a bigger overall market.
None of these risks are new. What changed is how much money now depends on them not materializing before the next round of earnings.
5 Predictions for the Second Half of 2026
Where this chip stock rally goes next depends on a handful of variables that should become clearer within two or three quarters.
- HBM pricing stays firm through year-end. With 2026 supply already sold out, Micron and SK Hynix have little reason to cut prices before new 2027 capacity comes online.
- Intel’s external foundry revenue becomes the number that matters most. Investors will watch whether Intel can convert customer conversations into signed wafer agreements, not just yield improvements on 18A.
- AMD’s Instinct GPU shipments face their real test. The MI-series roadmap needs to show actual revenue growth, not just design wins, to justify the stock’s re-rating.
- Nvidia responds with its own supply-chain moves. Expect Nvidia to lock in more memory capacity directly, extending its pattern of certifying and investing in suppliers like Samsung and SK Hynix.
- Volatility rises again before the next earnings cycle. With valuations stretched across the sector, any disappointing guidance from Micron, Intel, or AMD could trigger a sharp pullback even if the underlying AI demand story stays intact.
Frequently Asked Questions
Why did chip stocks rally so much in Q2 2026?
Investors broadened their AI infrastructure bets beyond Nvidia. Micron benefited from a sold-out HBM memory market, Intel from progress on its 18A manufacturing process plus a U.S. government equity stake, and AMD from record server CPU share and growing Instinct GPU demand. Together the three added about $2 trillion in market value, according to CNBC.
Which stock gained the most in the Q2 2026 chip rally?
Micron led with a 240% gain, adding roughly $920 billion in market capitalization. Intel followed at 216% (about $480 billion added), and AMD rose 186% (about $615 billion added).
Why didn’t Nvidia’s stock rally as much as Micron, Intel, and AMD?
Nvidia rose about 15% in the same quarter, far less than its peers, largely because its stock was already priced for AI dominance. Micron, Intel, and AMD had more room to re-rate because the market hadn’t fully priced in their roles in AI infrastructure.
What is HBM4 and why is it driving the memory shortage?
HBM4 is the newest generation of high-bandwidth memory, the specialized DRAM stacked next to AI accelerators to feed them data fast enough to keep the chips from sitting idle. Demand from AI data centers has outpaced supply, and Micron, Samsung, and SK Hynix have all reported their HBM output sold out under long-term contracts.
Does the U.S. government own part of Intel?
Yes. The U.S. government took a 10% equity stake in Intel in 2025 as part of an effort to keep advanced chip manufacturing onshore. That stake predates the 2026 chip stock rally but gave the stock additional support heading into it.
What is Intel’s 18A process node?
18A is Intel’s most advanced chip manufacturing process, central to its foundry turnaround strategy. Intel’s Panther Lake processors began shipping on 18A after launching at CES 2026, and the company has signaled growing openness to producing chips on 18A for outside customers, not just its own products.
Is the 2026 chip stock rally a bubble?
It’s too early to say. Some analysts point to elevated volatility and rapid momentum-driven gains as bubble warning signs. Others argue the rally reflects a real, multi-year memory and foundry shortage rather than pure speculation. Both can be true at once, which is why predictions for the second half of 2026 lean toward continued gains alongside sharper pullbacks.
What could slow down the AI chip rally in the second half of 2026?
Faster-than-expected HBM capacity from Samsung and SK Hynix, weaker-than-expected external foundry orders at Intel, or a disappointing earnings report from any of the three companies could all trigger a pullback. Broader AI infrastructure spending from hyperscalers like Microsoft, Google, Amazon, and Meta remains the biggest variable underneath all of it.
