Micron secures $22 billion in AI memory deals as chipmakers bet on breaking boom-bust cycle

Micron joins rivals pitching AI deals as cure for memory's boom-bust cycle

Memory chip giants like Micron are forging long-term "take-or-pay" deals, securing billions from customers like Nvidia to ensure consistent revenue. This strategy aims to break decades-old boom-bust cycles by guaranteeing cash flow even if the AI boom falters. These agreements, backed by customer cash commitments, signal a shift from treating memory as a commodity to a strategic necessity, potentially stabilizing the volatile industry. Memory chipmakers have for decades been trapped in boom-bust cycles, with capacity buildouts hitting the market just as demand craters. Micron, Samsung and SK Hynix are now trying to convince investors this time is different, arguing long-term deals will keep cash flowing even if the datacenter boom bursts. Micron said on Wednesday customers such as Nvidia had committed $22 billion to lock in supplies of memory chips, playing up huge growth in five-year "take-or-pay" deals that require clients to either buy its chips or hand over cash. The US company's deals follow in the ⁠footsteps ⁠of SK Hynix and Samsung, which have also been signing long-term supply agreements with their customers. The moves are key to winning over investors wary of the AI boom's durability, with memory stocks leading a $1 trillion-plus rout earlier this week stoked in part by valuation concerns. "The main question heading into Micron earnings... was how durable memory pricing power really is. What they showed, through longer-term strategic agreements is that visibility is improving and any downside risk is getting pushed further out," said Jake Behan, ETF-provider Direxion's capital markets head. "What matters from here is not whether memory pricing eventually normalizes as we know it likely will, it is about who captures ⁠and monetizes that pricing power while it lasts." Memory has become so critical to AI chips such as those made by Nvidia that customers no longer treat Boise, Idaho-based Micron as a commodity supplier to be played off rivals for ⁠lower prices, but as a strategic partner whose factory expansions they must underwrite to lock in supply. Despite joining the $1 trillion valuation club earlier this year, Micron reported an annual loss of $5.3 billion as recently as 2023, driven by a collapse in spending on consumer electronics after the frenzy of pandemic gadget upgrades. "Customers have put billions of dollars on Micron's balance sheet as a show of confidence and their commitment toward this new business model," the company's chief business officer, Sumit Sadana, told Reuters. Still, even with good-as-cash agreements in hand, Micron said it will take time for it to build out new factories, keeping supplies tight until at least 2027. Memory chipmakers have tried long-term deals before To be sure, the famously cyclical memory industry has tried to lock in long-term deals before. But past attempts failed to smooth ups and downs because memory was a commodity, letting electronics makers swap suppliers and squeeze prices at will. Even with AI, long-term hardware agreements could ⁠stand so long as customers see real demand and application. Any crack, whether a wobble in orders or doubts about the AI buildout, could send them back to the negotiating table. "The bear case is that these contracts only hold while supply remains tight. If demand softens and the market turns, there is a risk they are renegotiated or abandoned, which would quickly reintroduce volatility," said Ben Barringer, head of technology research at Quilter Cheviot. But this time things are different because there is real money on the line. Having customers pay cash to lock in commitments means Micron earns money regardless of whether those agreements go through or not. It also gives the broader AI demand narrative some legitimacy, showing that customers think it is worth spending billions just to ensure chip orders are confirmed.

Nvidia vs. Micron: The AI Stock Race Has A New 5-Year Winner - NVIDIA (NASDAQ:NVDA), Micron Technology (N

For most of the artificial intelligence boom, one stock has stood in for the entire trade: Nvidia Corp (NASDAQ:NVDA). The graphics-chip maker became the first company worth more than $4 trillion, and for nearly the whole of the past five years it sat at the top of any AI-era performance scorecard, including against its fellow semiconductor names. As recently as May 2026, a head-to-head of five-year returns still showed Nvidia comfortably in front. Then the leaderboard changed. A New Name At The Top The stock that passed it is Micron Technology Inc (NASDAQ:MU). On a five-year basis, Micron is now up roughly 1,320%, ahead of Nvidia's roughly 859%, according to TradingView data through June 30. For years the two charts told opposite stories: Nvidia climbed early and steadily as the data-center buildout took off, while Micron stayed flat, weighed down by the memory industry's familiar boom-bust rhythm. The crossover came almost entirely in the past two months, as Micron went near-vertical and Nvidia's stock cooled into a sideways range. Chart: Nvidia vs. Micron - 5Y Stock Price Performance Why Micron Did It The fundamentals explain the move. In its most recent quarter, Micron reported revenue of about $41.5 billion, up roughly 346% year-over-year -- a faster top-line acceleration than Nvidia, whose quarterly revenue grew about 85% over the same window. The gap is even starker on the bottom line. Micron posted adjusted earnings of $25.11 per share against Nvidia's $1.87, and its adjusted EPS grew more than 1,200% from a year earlier, versus Nvidia's still-formidable 131%. The engine is high-bandwidth memory, or HBM -- the fast memory stacked beside AI accelerators -- where surging demand and tight supply have driven prices, margins and earnings sharply higher. Cheap, And Micron Cheaper Still What makes the rally unusual is that, even after the run, neither stock looks expensive on forward earnings. Nvidia trades near 19.9 times next-twelve-month estimates -- below its 31.3x historical average -- while Micron sits around 8.0x, under its own 11.6x mean and well below Nvidia despite the bigger move. That discount reflects an old reflex: the market still prices memory as a cyclical commodity rather than as core AI infrastructure. The 'Memory Race' Thesis In a June 29 note, Jordi Visser - head of AI Macro Nexus research at 22V Research - wrote that "AI started as a compute race. It is becoming a memory race," citing KAIST's Kim Jung-ho -- the "father of HBM" -- who frames the GPU as a brilliant analyst and memory as the desk, filing cabinet and library that analyst needs to actually do the work. Micron has pointed the same way, tying future demand to AI agents, robotics and vehicles, and flagging long-term customer agreements worth roughly $100 billion through 2030. Whether memory keeps the lead is the open question, but the AI trade no longer has a single face. Market News and Data brought to you by Benzinga APIs To add Benzinga News as your preferred source on Google, click here.

'AI Equals Memory': Why The AI Boom Now Runs Through Micron, SK Hynix And Samsung - NVIDIA (NASDAQ:NVDA),

The defining bottleneck of the artificial-intelligence boom is no longer the graphics processor. It is the memory that sits next to it. That is the core argument of a Monday note from Jordi Visser, head of AI macro nexus at 22V Research. Borrowing a line from KAIST professor Kim Jung-ho -- the engineer often called the "father of HBM" -- Visser frames the whole trade in three words: "AI equals memory." His thesis is that the market keeps measuring the AI cycle by how far chip stocks have run, when it should be measuring how far the intelligence stack still has to evolve. "AI started as a compute race," Visser writes. "It is becoming a memory race." While compute created intelligence in the training era; in the agentic era, "memory gives intelligence context." From A Brilliant Brain to A Memory Palace Visser leans on Kim's analogy: the GPU is the brilliant analyst, but memory is the desk, the filing cabinet, the library and the courier system. A brilliant analyst with no filing cabinet and a slow courier spends the day waiting for files. Scale that to the millions of AI agents enterprises expect to run in parallel -- each one needing context, state and persistence -- and demand for memory does not taper off. It compounds. Micron Numbers Fedd the Narrative The earnings data is backing the story. Management guided fiscal fourth-quarter revenue to approximately $50 billion. CEO Sanjay Mehrotra said memory has become "a strategic asset" in the AI era and that "AI system performance is architecturally dependent on memory subsystem performance and capacity." More important for the cyclicality debate, Micron said it has now signed 16 strategic customer agreements -- multi-year deals with committed volumes and pricing terms. According to Visser, those contracts represent roughly $100 billion in cumulative revenue for Micron through 2030. Long-term supply agreements with committed pricing suggest customers increasingly view advanced memory as a strategic resource rather than a commodity component. A Market Dominated by Few Players Here is what makes the trade so concentrated. The number of firms that can supply AI-grade memory at scale is tiny. The bullish case is reinforced by the industry's highly concentrated structure. In DRAM, just three companies control roughly 90% of global production. According to Counterpoint Research, in the first quarter of 2026, Samsung Electronics Co. led with about a 38% revenue share, ahead of SK Hynix Inc. at roughly 29%, with Micron third. Goldman Sachs estimates SK Hynix has locked up about two-thirds of orders for Nvidia's next-generation HBM4. That HBM lead was strong enough to push SK Hynix past Samsung in annual operating profit for the first time in 2025. "From the charts alone, it is easy to say this must be late. But that is the wrong measuring stick. The better question is not how far the stocks have moved. The better question is how far the intelligence stack still has to evolve," Visser said. Industry Faces New Antitrust Challenge The concentration that underpins the bullish investment case is also attracting legal scrutiny. A proposed federal class-action lawsuit - Garciaguirre et al. v. Samsung Electronics Co., Ltd. et al. (No. 3:26-cv-06345) - filed on June 25 in the U.S. District Court for the Northern District of California accuses Samsung Electronics, SK Hynix and Micron of conspiring to restrict supply and inflate prices in the global DRAM market. The plaintiffs -- which include consumers, small businesses and repair companies -- allege the three manufacturers coordinated production cuts and pricing following the pandemic, artificially driving up memory prices despite slowing demand. They are seeking damages and injunctive relief under U.S. antitrust laws. Market News and Data brought to you by Benzinga APIs To add Benzinga News as your preferred source on Google, click here.

Samsung Electronics, SK hynix to benefit further from AI memory bottleneck - The Korea Times

Nvidia CEO Jensen Huang, center, leaves the SK hynix booth during the annual COMPUTEX expo in Taipei, Taiwan, June 2. Reuters-Yonhap Memory chips remain the key bottleneck in the artificial intelligence (AI) supply chain, with Samsung Electronics and SK hynix still positioned as strong beneficiaries of the cycle, Sanjeev Rana, head of Korea research at CLSA, said Tuesday. During a press briefing on the sidelines of the CITIC CLSA Northeast Asia Forum held at Conrad Seoul, Rana noted that over the past 10 years, memory chips' contribution to semiconductor industry revenue has jumped from 28 percent to 52 percent. "It's all driven by global AI infrastructure spending. Companies are looking to expand capacity across foundry, advanced packaging, production equipment, interconnects and other areas," Rana said. "But these bottlenecks are still likely to persist for the next several years and are unlikely to go away anytime soon." According to CLSA, global semiconductor industry revenue is expected to reach $2.5 trillion by 2030, up 80 percent year-on-year, with the majority -- approximately $1.4 trillion -- coming from the memory chip sector. Rana said that as AI models grow, limitations on memory bandwidth increasingly constrain overall system performance and scalability. In AI data centers, 60 percent of energy is spent on moving data across chips and only 40 percent on compute. While the industry has pursued multiple mitigation strategies, including faster memory, larger caches and improved data-movement architectures, the memory wall remains a persistent and costly challenge, he noted. The emergence of agentic AI, though still in its early stages, is also raising the need for CPUs, which consume a large amount of server DRAM. "In iPhones, Galaxy phones or PCs, companies in the past could lower memory content, and demand and supply would eventually come into balance," Rana said. "But if we try to lower the memory content in AI systems, it is going to have a very negative impact on the consumer experience and performance. So that kind of trade-off just isn't there." This outlook has prompted hyperscalers to sign multiyear supply contracts with Samsung Electronics and SK hynix. Rana forecast that more than 50 percent of their capacity will be tied to long-term agreements, providing better visibility for the companies' performance. CLSA has a target price of 540,000 won ($351.84) for Samsung Electronics and 3.7 million won for SK hynix. "The valuation multiple rerating story for Korean memory stocks is underway, and it will continue for some time," Rana said. Regarding the rise of Chinese memory chip makers, such as CXMT in DRAM and YMTC in NAND, Rana said he does not expect them to compete directly with Samsung and SK hynix in the near term, as the performance and power efficiency of their chips remain significantly lower. "But it is also a fact that Samsung and SK hynix can meet only 60 percent of customer demand. For the remaining 40 percent, Chinese memory chip makers do have opportunities," he said, noting that Apple CEO Tim Cook has hinted at easing restrictions on Chinese suppliers.

Micron Re-Rating Tests Whether AI Memory Has Broken the Old Cycle

Micron (MU) delivered the loudest earnings print of the chip season and still couldn't hold the gains. The stock traded near $1,132 into Monday after a violent two-way week: it ripped 15% to 17% on a blowout fiscal third quarter, then got smoked in the broader memory-sector selloff that dragged the complex lower into the weekend. The result is a name that touched roughly $1,213 post-earnings before the bid faded, leaving it consolidating well off its highs but still up an astonishing amount on the year -- a $10,000 stake at the start of 2026 is worth more than $30,000 today. The structural story is what makes Micron the most-watched memory name on the tape. The company has broken free from its historical identity as a purely cyclical, low-margin commodity producer, re-rated by the unprecedented shift toward High-Bandwidth Memory and AI-optimized chips. The GPUs powering the AI buildout need enormous quantities of HBM, the specialized, high-margin product Micron makes, and that single shift turned the stock from a textbook cyclical into one of the hottest large-caps on the market. Micron crossed a $1 trillion market capitalization on May 26, 2026, the latest US name to join the club on surging HBM demand. The fiscal Q3 numbers validated the re-rating in full. Revenue hit $41.5 billion, up 346% from the year-ago quarter, with adjusted earnings of $25.11 per share crushing the $20.78 the desk modeled. Net income reached $28.2 billion and the profit margin expanded to 68% from 20% a year earlier. The Q4 outlook called for roughly $50 billion in revenue, sequential growth that few legacy semiconductor names could dream of posting. The thesis for this forecast is direct: Micron is a structurally re-rated AI-memory leader trading at a deceptively modest multiple, but the stock carries the ghost of its cyclical past and got caught in last week's chip-sector unwind. The bull case rests on sold-out HBM capacity, 81% gross margins, and an average analyst target near $1,410. The bear case is the capex surge and the cyclical risk that haunts every memory name. The post-earnings consolidation near $1,132 is the battleground -- hold the breakout zone and the re-rating extends; lose it and the cyclical ghost gets a louder voice. The Price Scoreboard: A Tenfold Run and a Whipsaw Week The 52-week range tells the whole story of the re-rating. Micron has traded from a low near $103 to an all-time high above $1,089, a roughly tenfold run that ranks among the most explosive moves in the large-cap universe over the past year. The stock is up between 591% and 830% over various trailing-year windows depending on the start point, having tracked significantly ahead of even its blistering earnings growth -- share price compounding near 166% annually over three years against earnings-per-share growth of 131%. The recent action has been violent in both directions. Heading into the fiscal Q3 report, the stock hit an all-time high near $1,051 and closed one session at $1,051.77 before the memory complex rolled over. During the sector rout, Micron tumbled 13% in a single session, then rebounded 4.1% the next day -- the kind of whipsaw that defines a crowded, high-beta name. The earnings beat sent it ripping 15% to 17% on the print, pushing it toward $1,213, before the Friday selloff in the memory group knocked it back toward $1,132, a roughly 6.7% single-session give-back. The pullback was sector-wide, not company-specific. Micron, SanDisk, and Western Digital all fell together in the session after the blowout, with the memory and storage group rolling over as the broader AI trade cooled, an OpenAI IPO-delay report rattled the chip space, and the semiconductor index slid toward correction territory down nearly 10%. The flagship semiconductor ETF shed more than 5% on the week, and the chip bellwether entered its own correction -- the kind of macro-driven group move that drags the strongest names down with the weakest. The setup heading into the week is a post-earnings consolidation. The stock has the fundamental wind of a record quarter at its back but the technical headwind of a sector that just got smoked. Near $1,132, Micron sits below its post-earnings high near $1,213 and its all-time high near $1,089 on an intraday basis -- the levels reshuffled by the violent swings. The price action is the textbook profile of a name digesting a monster print inside a hostile sector tape, where the next move depends on whether the AI-memory bid reasserts or the cyclical sellers press their case. The Blowout Quarter: A Record That Reset the Model The fiscal third-quarter print didn't just beat -- it reset the entire earnings model. Revenue of $41.5 billion marked a 346% surge from the year-ago quarter, blowing past the roughly $35.75 billion the desk had penciled in. Adjusted earnings of $25.11 per share topped the $20.78 consensus by a wide margin and stood against just $1.69 in the comparable quarter a year earlier -- a near-fifteenfold year-over-year jump that captures how completely the memory pricing environment has flipped. The profitability metrics are what stopped the desk cold. Net income reached $28.2 billion, up roughly $26.4 billion from the year-ago period, and the profit margin expanded to 68% from 20% a year earlier. For a company historically defined by thin, cyclical margins that compressed toward breakeven during every downturn, a 68% net margin is a structural anomaly -- the direct result of tight DRAM and NAND supply colliding with surging AI demand to hand Micron pricing power it has never held before. The guidance extended the beat rather than capping it. The company pointed to roughly $50 billion in revenue for the fiscal fourth quarter, sequential growth of about 20% off an already-record base, and guided gross margins toward 81% -- extraordinary for a memory maker. That margin guide is the number that separates this cycle from every prior one: when a commodity producer guides to 81% gross margins, the market has to decide whether it's a temporary shortage premium or a durable structural shift in the product mix toward high-value HBM. The market's reaction split the difference. The stock initially ripped 15% to 17% on the print as the headline numbers validated the AI-memory thesis, with the earnings and guidance sending the entire computer memory and storage group higher in sympathy. But the focus quickly shifted from the strength of the current quarter to the sustainability of the pricing, margins, and the memory shortage driving it -- the perennial question that haunts every memory cycle. The pullback that followed wasn't a rejection of the numbers; it was the market wrestling with whether numbers this good can possibly last, and that tension defines the stock's near-term path. The HBM Supercycle: Sold Out and Locked In The engine behind the re-rating is High-Bandwidth Memory, and the supply-demand dynamics there are unlike anything in Micron's history. The company has reportedly sold out its entire 2026 HBM production and is locking customers into long-term supply agreements stretching up to five years, into 2027 and beyond. When a chipmaker's output is spoken for a year or more in advance, pricing power follows automatically -- the customer competes for allocation rather than negotiating on price, which is the inverse of the commodity dynamic that defined memory for decades. The margin math flows directly from that scarcity. Micron guided toward gross margins near 81% for the upcoming quarter, a figure that would have been unthinkable for a memory maker even two years ago. Tight DRAM and NAND supply plus the surging AI demand have flipped the pricing equation decisively in Micron's favor, and the sold-out HBM book means that pricing holds through the contract terms rather than collapsing at the first sign of a demand wobble. The long-term agreements convert what was a spot-priced commodity into something closer to a contracted, recurring-revenue business. The technology roadmap reinforces the moat. Micron is ramping HBM4 production with its multi-generation leadership in DRAM and NAND process technology, most recently its 1-gamma DRAM and G9 NAND nodes. The chip bellwether has certified Micron alongside Samsung and SK Hynix to supply HBM4 for its next-generation AI platform, reinforcing Micron's role at the high end of the AI memory stack. The company has also been first to market with advanced products -- sampling 256GB DDR5 server modules and shipping industry-leading 245TB data-center SSDs -- that extend its reach across the AI infrastructure footprint. The structural question is whether the shortage persists. Supply increases may not arrive until mid-2027, which could keep the memory market's supply-demand dynamics in Micron's favor through at least fiscal 2027. The company's own massive capacity additions are not expected to add meaningful supply until fiscal 2028, meaning the tight conditions that drove the record quarter have a multi-year runway before new capacity floods the market. That timeline is the bull case's foundation: a sold-out book, contracted pricing, and a supply response that's structurally delayed gives Micron a window of extraordinary profitability that the cyclical skeptics have consistently underestimated. The AI Demand Engine: Hyperscalers Pay Up The demand side of the equation is being driven by an AI buildout that treats memory as a non-negotiable input. Big Tech companies are willing to pay astronomical prices for AI memory components, helping spark a dramatic turnaround in Micron's finances -- what one observer described as an extraordinary transfer of cash from the providers of AI to the memory-chip makers. The memory shortage has become acute enough that it's reshaping the economics of the entire device ecosystem. The shortage's reach extends to the largest names in tech. The squeeze has been described as an existential crisis for smaller players, and even the giants are feeling it -- one consumer-hardware leader raised MacBook and iPad prices on the back of rising memory costs and has been seeking a China chip deal to ease the AI-driven component inflation. When the most powerful hardware companies in the world are raising end prices and scrambling for supply, the pricing power sits unambiguously with the memory makers, and Micron is the only major American producer in the Big Three. The strategic partnerships lock in the demand. Micron struck a multi-year AI infrastructure agreement to become a primary supplier of memory and storage for next-generation AI workloads, co-designing HBM and storage subsystems with the partner and participating in its funding round. The company is deepening technical collaboration across the AI ecosystem through cooperative design, bringing platforms to market faster with greater system-level optimization. These aren't spot-market sales -- they're embedded, co-engineered relationships that tie Micron's roadmap to the hyperscaler buildout for years. The geographic and capacity footprint underpins the ability to deliver. Micron is backing these commitments with major manufacturing investments across the US, India, Japan, Singapore, and Taiwan, positioning it to scale the innovations as demand ramps. The breadth of the demand -- spanning cloud, core data center, mobile, and client segments -- means the company isn't dependent on a single end market. The AI engine driving the memory shortage shows no sign of cooling through the contracted window, and as long as the hyperscalers keep paying up for HBM, Micron's pricing power holds. The Capex Question: The Bear's Cyclical Ghost The single largest source of the post-earnings hesitation was capital spending, and it's where the cyclical ghost lives. Micron raised its investment plan for fiscal 2026 to more than $25 billion, with roughly $7 billion deployed in the third quarter alone, and warned of further spending increases into 2027. The market interpreted this as an acceleration in capacity expansion and a potential normalization of the cycle down the road -- the classic memory-industry pattern where today's shortage-driven profits fund tomorrow's oversupply. The scale of the long-term commitment is staggering. The company has committed over $100 billion to new fabs as part of a planned $200 billion US expansion, including a megafab project in New York built with a major construction partner and advanced 1-alpha DRAM production already underway at its Manassas, Virginia facility. These are generational investments that tie Micron into the long-term AI and hyperscaler demand agreements, but they also represent the kind of capacity buildout that has historically preceded the down leg of every memory cycle. The competitive capex adds to the worry. The pressure isn't only internal -- a Korean rival placed an order of nearly $8 billion with the dominant chip-equipment maker to expand its own HBM and DRAM capacity, signaling that the entire industry is racing to add supply into the AI boom. When every major producer is simultaneously expanding, the seeds of the next oversupply get planted during the peak of the shortage, which is precisely the dynamic that has made memory stocks treacherous to own through a full cycle. The bull's rebuttal is timing. The crucial detail is that this new capacity is not expected to add meaningful supply until fiscal 2028, which preserves the tight supply-demand dynamics through at least fiscal 2027. The capex is funding a structurally different product -- high-value HBM tied to long-term contracts -- rather than the commodity DRAM that flooded prior cycles. The key tension for the stock is exactly this contrast: very strong current fundamentals backed by long-term contracts against a memory market that still carries a history of sharp cycles. Whether the capex surge signals prudent expansion or the prelude to normalization is the question the bears and bulls are fighting over. The Fundamentals: A Business Transformed The segment structure shows a company built around the AI opportunity. Micron operates through its Cloud Memory Business Unit, Core Data Center Business Unit, Mobile and Client Business Unit, and its broader memory and storage portfolio, delivering DRAM, NAND, and NOR products under the Micron and Crucial brands. The data-center-oriented segments are where the HBM and high-value DRAM demand concentrates, and they've driven the bulk of the revenue surge as the product mix shifts toward AI infrastructure. The full-year forecasts have been revised upward dramatically. The consensus outlook for fiscal 2026 revenue jumped from $79.8 billion to $108.7 billion, while the full-year adjusted EPS estimate climbed from $34.26 to roughly $58 per share -- against just $7.65 in the prior fiscal year. Net income is forecast to grow 251% next year, versus a 38% growth forecast for the broader US semiconductor industry, underscoring how far Micron's trajectory has decoupled from the sector average. The growth runway extends well beyond the current year. Revenue is forecast to grow 29% annually on average over the next three years, compared to a 23% growth forecast for the US semiconductor industry -- a premium growth rate sustained by the contracted HBM book and the multi-year AI agreements. The company also pays a modest dividend, declaring a quarterly payout of $0.15 per share with an ex-date of July 6 and payment on July 21, a token return that signals confidence in the cash-flow durability without diverting capital from the expansion. The cash-generation transformation is the deepest part of the story. A company that earned $7.65 per share in a recent fiscal year is now tracking toward roughly $58 -- a step-change that reflects both the volume ramp and the margin expansion to 68% net. The vertical integration and strong product pipeline support the case for sustained high margins, with significant improvements in AI inference and CPU demand combining with constrained supply to drive continued pricing increases. The fundamental picture is of a business that has genuinely transformed its earnings power, leaving the open question of how long the supercycle conditions persist rather than whether they're real. Valuation: Cheap on Paper, Cyclical in Practice The valuation paradox is what makes Micron so polarizing. Despite the tenfold run, the stock trades at a fiscal-2026 third-quarter P/E of roughly 20.6x and an EV/EBITDA of about 15.0x -- multiples that look almost value-like for a company growing earnings at triple-digit rates. On the surface, a 20x multiple on a business compounding earnings this fast screams cheap, which is the core of the bull's argument that the stock has room to run even after its enormous move. The catch is the denominator. Memory earnings are notoriously cyclical, and a low P/E on peak-cycle earnings can be a value trap rather than a bargain -- the multiple compresses precisely because the market assumes the earnings will mean-revert. The forward valuation looks even more compelling at face value: against the roughly $58 EPS estimate, the stock trades at a single-digit forward multiple, which is why some analysts anchor targets to a 9x multiple on forward earnings. But that math only holds if the earnings prove durable rather than peaking. The market-cap context frames the stakes. Micron crossed $1 trillion in market capitalization on May 26, joining the most exclusive tier of US companies, and the question now is whether a memory maker can sustain a trillion-dollar valuation through a full cycle. The dispersion in analyst price targets captures the uncertainty vividly -- the range spans from a low near $361, implying roughly 68% downside, to a high near $2,200, implying roughly 94% upside. That spread is among the widest on any large-cap, and it reflects genuine disagreement about whether the re-rating is structural or cyclical. The valuation verdict hinges on the cycle's durability. If the HBM supercycle and the supply tightness persist through fiscal 2027 as the bulls expect, the current multiple is genuinely cheap and the stock re-rates higher toward the $1,410 average target. If the capex surge normalizes the cycle and demand cools after 2027, the peak earnings compress and the low multiple proves a warning rather than an opportunity. The stock is cheap on paper and cyclical in practice, and which characterization wins depends entirely on whether this cycle is different from every memory cycle that came before it. The Technical Structure: Digesting the Spike The technical picture is a name consolidating a violent earnings spike inside a hostile sector tape. After ripping toward $1,213 on the print, the stock pulled back to near $1,132 as the memory group rolled over, leaving it in a digestion phase below its post-earnings high. The all-time high near $1,089 on a prior basis and the intraday spike high near $1,213 frame the upper boundary, while the pre-earnings consolidation around $1,000 to $1,051 marks the zone the stock would need to defend on any deeper pullback. The momentum profile reflects the whipsaw. The stock swung from a 13% single-session drop during the sector rout to a 15%-to-17% earnings rip and then a 6.7% give-back -- the kind of elevated volatility that options markets had priced in ahead of the report. That violent two-way action leaves the shorter-term indicators choppy and the trend unresolved, with the stock caught between the fundamental tailwind of the record quarter and the technical headwind of a semiconductor index that just slid toward correction. The sector backdrop is the dominant technical force. The flagship semiconductor ETF shed more than 5% on the week, the chip index verged on a 10% correction, and the bellwether name entered its own correction -- a group-wide derisking that overwhelmed Micron's company-specific strength. When the entire complex is selling off on macro AI-spending fears and an IPO-delay headline, even a blowout quarter struggles to hold its gains, and Micron's pullback was as much about the sector as about the stock. The structure resolves at the breakout zone. The level that matters most near term is whether Micron can hold above its pre-earnings consolidation and the $1,000-to-$1,050 base that supported it into the print. A hold there keeps the post-earnings breakout intact and positions the stock to retest its highs once the sector stabilizes. A break below it would signal the earnings spike has fully reversed and the cyclical sellers have regained control, opening a deeper retracement. The technical bias is constructive as long as the base holds, but the sector tape has to cooperate for the breakout to extend. The Downside Map: $1,050, Then the Base The support structure beneath the spot is defined by the pre-earnings consolidation. The first line sits at the $1,050 zone, near the all-time-high closing level around $1,051.77 that the stock defended into the report. That level has flipped from resistance to support after the earnings breakout, and holding it would confirm the spike has consolidated into a higher base rather than reversing. A daily close below it would be the first sign the post-earnings strength is fading. Below $1,050, the next shelf is the $1,000 round number and the broader pre-earnings base. The stock built a consolidation in the high-$900s to $1,000 range ahead of the print, with the explosive surge that pushed it toward the top of an $895-to-$997 daily range marking the launch point. That zone represents the structural support where the breakout originated, and a retreat into it would test whether the buyers who drove the pre-earnings accumulation step back in. A hold there preserves the longer-term uptrend; a break opens a deeper unwind. The deeper downside targets reflect the sector and cyclical risk. If the memory-group selloff intensifies and the cyclical narrative gains traction, the stock could retrace toward the high-$800s, the level where one analyst snapshot pegged a recent close near $896 before the latest leg higher. The wide analyst target range, with a low near $361, captures the tail risk that a full cyclical reversal would imply -- though that scenario requires the supercycle thesis to break entirely, which the sold-out HBM book and contracted pricing argue against in the near term. For the forecast, the downside hinges on the $1,000-to-$1,050 base. As long as that zone holds on a closing basis, the post-earnings breakout stays intact and the stock is consolidating rather than reversing. A confirmed break shifts the framework: the high-$800s come into play, the cyclical sellers gain the upper hand, and the burden of proof shifts back to the bulls. The desk should treat the pre-earnings base as the pivot -- the area that determines whether the record quarter marked a new floor or a blow-off top. The Upside Map: Reclaiming $1,213 and the Path to Target The resistance structure above the spot starts with the post-earnings spike high. The first hurdle is the roughly $1,213 level the stock touched on the earnings rip before the sector selloff knocked it back. Reclaiming that high on rising volume would signal the AI-memory bid has reasserted and the breakout is resuming, clearing the path toward new all-time highs. Until the stock takes out that spike high, the post-earnings move remains incomplete and vulnerable to further consolidation. The path beyond the spike high leads toward the analyst targets. The average 12-month price target sits near $1,410, implying roughly 23% to 25% upside from current levels, with the Strong Buy consensus from 43 analysts reflecting broad conviction in the structural story. A break above $1,213 and a sustained move higher would put that $1,410 zone in play, and the most aggressive targets stretch far beyond -- one firm raised its target to $1,870 post-earnings, and the highest target on the Street sits near $2,200. The catalyst stack supports the upside case. The post-earnings analyst revisions have been overwhelmingly positive, with multiple firms raising targets sharply -- one major bank lifted its target by 70%, another doubled its target, and a leading shop chose Micron as a top AI name to own. The sold-out HBM capacity, the 81% margin guide, and the $108.7 billion full-year revenue forecast give the bulls concrete numbers to anchor to, and the HBM4 ramp into the next-generation AI platforms provides a forward catalyst that extends the demand visibility. The upside requires the sector to cooperate. The mechanism for a move to new highs exists -- the fundamental case is intact and the targets sit well above spot -- but the stock needs the broader semiconductor complex to stabilize for the breakout to extend. As long as the chip index is sliding toward correction and the AI trade is being questioned, even Micron's blowout numbers struggle to drive the stock to new highs. The path to the $1,410 target runs through a reclaim of $1,213 and a sector tape that stops fighting the AI-memory bid. Once both align, the upside opens; until they do, the resistance overhead caps the rallies.

AI's Hidden Cost Problem Is Making Memory Chip Makers the Biggest Winners

But markets do not hand out equal portions at the buffet. Somewhere in every boom, a bottleneck appears. The question is simply which clock starts ticking first. For now, the memory makers are sitting at the head of the table with the champagne bottle. The hyperscalers are paying for dinner, the model makers are giving away dessert and the end users are still asking whether the meal is worth the price. Takeaways * AI is fragmenting from a broad "buy everything" theme into a much narrower fight over who controls the bottlenecks. * Memory suppliers are currently behaving like oil producers during an airline boom: essential, scarce and suddenly able to dictate terms. * The hyperscalers and model providers are absorbing rising costs because the strategic race matters more than near-term margin discipline. * The eventual pressure valve comes from efficiency, monetisation or new supply. Until one of those arrives, the companies controlling memory remain the ones collecting the toll. AI's Hidden Cost Problem The AI boom is beginning to look less like a clean technological revolution and more like an industrial boomtown where everyone arrived chasing gold, only to discover the man selling water owns the whole river. For two years, investors have treated AI as one giant winning ticket. Buy the chips. Buy the cloud. Buy the power names. Buy the data-centre landlords. Buy anything with a server rack in the basement and a slide deck containing the words "agentic workflow." The assumption was simple: the whole AI food chain would feast together. But markets do not hand out equal portions at the buffet. Somewhere in every boom, a bottleneck appears. Somebody owns it. And once the room realises there is only one narrow door out, that owner starts charging rent. Right now, that door is memory. Micron, Samsung and SK Hynix are becoming the oil producers of the AI era. The hyperscalers are the airlines, the model makers are the travel agents, the users are still booking discount seats and the memory suppliers are standing at the fuel depot with the hose in one hand and the invoice in the other. High-bandwidth memory is not a decorative extra. It is the oxygen tank strapped to the back of the AI machine. The models need it to train, infer, scale and keep responding without coughing up smoke. And because capacity is thin, qualification cycles are long and fabs take years rather than quarters to build, the market has gone from orderly to feral. Prices are climbing sharply while shipments barely move. That is never a normal demand story. That is the market equivalent of a crowded bar during a blackout: nobody is drinking more, but the bartender has doubled the price of every bottle because there are only three candles left. The result is an enormous transfer of cash. The memory suppliers are not merely selling more chips. They are extracting a larger share of every AI dollar being spent. The cost is rolling uphill into the hyperscaler capex budgets, sideways into model-provider losses and eventually, perhaps, toward the end user. But for now, nobody wants to be the first to pass the bill on. That is the awkward part of the AI story. The technology may be extraordinary, but the economics are still wearing training wheels. Model providers are still pricing for adoption. Enterprises are still experimenting. Corporate users are burning through tokens like tourists spending casino chips because someone else is picking up the tab. The AI platforms are still more interested in filling the stadium than collecting enough at the gate to pay for the lights. So when memory costs go vertical, the bill does not disappear. It simply lands in somebody else's lap. For consumer electronics, the answer is easy enough. Raise the price of the laptop, watch buyers hesitate, clear the shelves and wait for the cycle to cool. Consumers can delay a purchase. They can keep the old machine. They can decide that last year's laptop is suddenly not that slow after all. But AI is not being run like a normal consumer-electronics cycle. The hyperscalers cannot simply shrug and pause the buildout. That would be like dropping your shovel in the middle of a gold rush because the price of diesel went up. Nobody wants to be the company that blinked while everyone else kept building. Nobody wants to explain to investors that they throttled capacity just as rivals were hoovering up enterprise workloads, developer ecosystems and the next generation of AI agents. So they absorb the cost. They can afford to, at least for now. The balance sheets are large enough. The cash flows are deep enough. The strategic stakes are high enough. But the market is beginning to see the difference between the companies selling the shovels and the companies digging the hole. That distinction matters. The memory names have been running like they have found a trapdoor beneath the AI casino. Meanwhile, parts of the hyperscaler complex have begun to look heavy. The market is starting to ask a harder question: who is actually making the money from this buildout, and who is simply writing the cheques? That is where the AI trade gets more interesting. For a while, the market was happy to reward the entire ecosystem. Nvidia, chips, cloud, data centres, utilities, cooling, cables, turbines, anything with a pulse. It was a rising-tide trade, and every boat was floating whether it had an engine or not. Now the tide is still in, but the water is being diverted. The memory suppliers are capturing more of the profit pool because they control the scarce input. They are the toll booth on the only highway into town. And when the highway is jammed with hyperscalers racing to build the future, the toll booth becomes a very profitable place to stand. That does not mean the trade is risk-free. Memory has a long and brutal history. It is the sort of business that can look like a money-printing machine right up until the day supply catches up and the whole thing turns into a yard sale. High prices attract capacity. Fat margins attract ambition. And when new fabs finally arrive, the industry can move from champagne to cold takeaway beer with breathtaking speed. The bulls will argue this time is different. There are fewer major players. The technology is more complex. High-bandwidth memory is not just another commodity component rolling off a production line. That may be true. But when a stock becomes the hottest table in the casino, the danger is not that the story is wrong. The danger is that everyone already knows the story, everyone owns the story and half the room has borrowed money to own more of the story. The larger point is that AI is no longer one trade. It is becoming a series of competing profit pools, each trying to pass the cost further down the chain. The chip makers are charging more. The hyperscalers are funding the buildout. The model providers are subsidising adoption. The end users are still deciding whether the product is useful enough to pay real money for it. Somewhere in that chain, the economics have to settle. There are only three ways this plays out. The first is that the buyers swallow the cost and margins take the hit. That is the immediate answer. The second is that the industry gets smarter: smaller models, better optimisation, lower memory intensity, tighter token budgets and less waste. That is the long-term answer. The third is that supply eventually arrives, the shortage loosens and the memory suppliers lose some of their pricing power. All three will happen. The question is simply which clock starts ticking first. For now, the memory makers are sitting at the head of the table with the champagne bottle. The hyperscalers are paying for dinner, the model makers are giving away dessert and the end users are still asking whether the meal is worth the price.

The Price of High Margins

Since the start of the AI rally, investors have been enamored with the ever-expanding margins of the few companies at the heart of this revolution. However, they are beginning to see the flip side of the coin: someone will have to foot the bill. Last Wednesday, Micron reported earnings that significantly beat expectations, fueled by a shortage of memory chips. In the last quarter, the company posted a gross margin of... 85%! The following day, Apple announced price increases for iPads and MacBooks due to this memory chip shortage. Price hikes ranging from 15 to 25% depending on the model could weigh on demand. On the same day, Microsoft announced it was raising the price of its Xbox consoles by $100 to $150. Both stocks fell following these announcements. Beyond the impact on sales, rising costs are already forcing hyperscalers to constantly revise their capital expenditures upward, a trend the market is no longer buying into. Microsoft, Meta, Alphabet, and Amazon have seen declines of 10 to 20% since the beginning of the month. Over the same period, Micron has gained 17%. When they released their first-quarter results in late April, the hyperscalers all raised their Capex forecasts. They are expected to collectively spend $725bn in 2026 on their AI infrastructure. According to Morgan Stanley, this figure is expected to climb to $1000bn next year. This investment frenzy and doubts regarding their profitability are being felt in stock prices. The valuation premium of the Magnificent 7 relative to the S&P 500 is at its lowest level in 10 years. Source: Apollo Global Management The big question now is whether or not hyperscalers will decide to slow down their investments. A brake that is too sudden would be perceived as a challenge to the AI theme and would be penalized, but a downward adjustment could instead signal a return to financial discipline and thus be rewarded. Ultimately, if AI becomes more expensive, it could slow its own development. As the Wall Street Journal explained last month, after an initial phase of unbridled AI use, companies are increasingly restricting usage in the face of skyrocketing costs. A Still-Cyclical Industry? Historically, the memory chip market is cyclical. During periods of high demand, prices rise and manufacturers invest to produce more. However, these investments only bear fruit several months or even years later. If demand weakens in the meantime, the market finds itself with excess production capacity. This oversupply then drives prices down, and manufacturers stop investing. Just after Micron's results were published, Chief Business Officer Sumit Sadana found it convenient to blame his customers for the current price increases. "We let some customers who were being very aggressive on pricing at the time know that this attitude was not constructive," he said in an interview with the Wall Street Journal. "Many industry investments were abandoned in 2023 due to very poor price and margin levels." AI brings hope that the industry may no longer be subject to this cyclicity. Supply is expected to exceed demand for several more quarters. Analysts also point out that memory chip suppliers are now signing more long-term contracts. Nevertheless, valuation levels remain very low, a sign that investors are not entirely buying into a change in the sector's status. Despite an 800% surge over one year, Micron is trading at only 15 times earnings. Current margin levels therefore do not seem sustainable over time. Not only because high prices affect demand, but also because they incentivize increased production. This morning, Samsung and SK Hynix announced they are investing $880bn in South Korea to increase their production capacity.

Memory chip makers reap AI windfall as prices surge, WSJ reports By Investing.com

Investing.com -- Memory chip makers are capturing an increasing share of profits from the artificial intelligence boom as soaring demand for high-bandwidth memory (HBM) pushes up prices and raises costs for AI developers and cloud providers, the Wall Street Journal reported on Saturday. Micron Technology said prices for its DRAM memory chips rose more than 60% in the quarter ended May 28 compared with the previous quarter, while NAND flash memory prices increased more than 80%. Shipment volumes rose only modestly, indicating that much of the revenue growth was driven by higher prices. HBM chips have become a critical component for AI servers and data centers, but limited manufacturing capacity has kept supply tight as demand continues to grow. The global HBM market is dominated by three manufacturers: Micron, Samsung Electronics, and SK Hynix. Building additional production capacity requires significant investment and can take several years, limiting near-term supply growth. Higher memory prices are increasing costs across the technology sector. Apple this week raised prices for several MacBook and iPad models, citing rising memory costs, while AI developers continue to face higher infrastructure expenses. Many AI companies have so far absorbed those costs rather than passing them on to customers, as firms continue competing to expand their user bases and market share. The increased spending has also shifted investor attention toward memory manufacturers. Shares of Micron, SK Hynix, and Samsung have outperformed several major technology companies this year as demand for AI infrastructure continues to accelerate. Cloud providers and AI model developers, including Microsoft, Amazon, Alphabet, and Meta Platforms, continue investing heavily in AI infrastructure despite the higher component costs. Additional memory capacity is expected to come online over the next several years, though current supply constraints are likely to persist until new manufacturing facilities begin production. The report said rising memory prices have highlighted the growing importance of semiconductor suppliers within the AI supply chain, as chip manufacturers benefit from one of the strongest pricing environments the industry has experienced in years.

AI Is Starving the Rest of the Memory Chip Market

In an industry long defined by brutal cycles of glut and shortage, Deutsche Bank believes memory is moving into a more structural phase, because artificial intelligence is not only consuming more capacity, it is reshaping the customer pecking order. Hyperscalers are locking up High Bandwidth Memory (HBM) with long-term contracts and are willing to pay premiums to secure supply, pushing Samsung, SK Hynix, and Micron to reallocate capacity toward the highest-margin products. Against that backdrop, the bank expects average annual HBM demand growth of about 40% through 2030, versus 21% for standard DRAM. This shift matters even more because HBM does not simply layer neatly onto the rest of the memory market, it partially crowds it out. Deutsche Bank notes that producing one additional bit of HBM requires roughly three times more silicon than a bit of conventional DRAM, a ratio that could rise further with HBM4 and HBM4e generations. As a result, the more manufacturers prioritize AI-bound chips, the tighter available supply becomes for PCs, smartphones, traditional servers, industrial equipment, and cars. That allocation shock toward HBM is already showing up in pricing. Deutsche Bank points to contract increases of 58% to 63% for standard DRAM in the second quarter of 2026, and 70% to 75% for NAND. PC makers such as Lenovo, Dell, and Asus have warned of potential price increases of 15% to 20%, while the added cost in autos could reach $150 to $600 per vehicle depending on the level of equipment. Memory is therefore becoming a macroeconomic issue, as it represents a serious source of durable-goods inflation. The bullish case for memory makers is not without risk, however. In 2026, multi-year contracts, capacity discipline, and hyperscaler demand still strongly support their pricing power. But the equation could become more unstable starting in 2027, when some incremental capacity begins to arrive and Chinese manufacturers relieve part of the entry-level segments. At the same time, software-optimization efforts could emerge to reduce memory consumption per query.