The Day Micron Joined the Trillion-Dollar Club
Micron Technology crossed the $1 trillion market capitalisation mark for the first time on May 26, 2026, becoming only the second memory company in history to achieve the milestone after Samsung. The catalyst was a UBS upgrade that nearly tripled the firm's price target for Micron — from $535 to $1,625, the highest among the 46 brokerages tracking the company — sending shares surging 18% to close at approximately $891. The reasoning behind the upgrade was straightforward: high-bandwidth memory chips, the kind Micron makes, have become the single most critical bottleneck in the AI infrastructure buildout, and right now Micron cannot make them fast enough. Micron's HBM capacity is fully sold out through the end of 2026 under long-term agreements, with next-generation HBM4 production already ramping for 2027 delivery. The stock's rise marks one of the fastest in semiconductor history, with Micron gaining more than 200% in 2026 alone and climbing over eight times from its lows during the memory-chip downturn just a few years ago. UBS estimates Micron could generate more than $400 billion in free cash flow between 2027 and 2029 — a number that, if achieved, would make the $1 trillion valuation look conservative in retrospect.
The broader market context of May 26 amplified Micron's significance. The S&P 500 and Nasdaq Composite both rose to fresh intraday all-time highs on the day, driven by technology sector strength and growing optimism about a potential U.S.-Iran ceasefire deal being reached this weekend. The Dow Jones Industrial Average fell 118 points on the same day — a divergence that captured the structural bifurcation in the U.S. equity market: AI-exposed technology companies hitting records while traditional industrial and financial companies reflected slower economic conditions. Micron's $1 trillion crossing was the headline event of a trading day that also saw Treasury yields fall as bond markets returned from the Memorial Day holiday, creating an unusual combination of equity euphoria, bond relief, and geopolitical optimism that market historians will likely mark as one of the more remarkable single sessions of the AI era.
What High-Bandwidth Memory Is and Why It Is Running Out
High-bandwidth memory is a specialised chip architecture that stacks multiple layers of DRAM directly on top of or adjacent to GPU and AI accelerator chips, connected through a high-speed interface that delivers data to the processor at rates far exceeding conventional memory architectures. Where standard DDR5 DRAM delivers memory bandwidth of approximately 100 gigabytes per second, HBM3E — Micron's current primary AI product — delivers over 1.2 terabytes per second. For AI inference and training workloads, which require feeding enormous models with data at the highest possible rate to generate tokens or complete training passes efficiently, memory bandwidth is as important as raw compute performance. An Nvidia H100 or B200 GPU paired with insufficient HBM bandwidth is like a sports car with a clogged fuel line: the engine is capable but the performance is throttled by the supply constraint. The global AI infrastructure buildout — which Nvidia's record $75.2 billion quarterly data centre revenue confirms is proceeding at an extraordinary pace — creates demand for HBM that is growing faster than either Micron, SK Hynix, or Samsung can expand production capacity. Data centres are on track to consume 70% of all memory chips manufactured this year, with memory manufacturers reportedly selling their 2028 capacity now.
The HBM supply constraint is structural rather than cyclical, which is the core of UBS's investment thesis. Expanding HBM production capacity requires building new fabrication facilities with the advanced node processes and packaging technology required for HBM's stacked architecture. These facilities take 18 to 24 months to plan, permit, and construct, and then additional months to ramp to full production yield. The AI demand that is consuming current HBM supply has been growing at rates that no production expansion programme can match in real time. Micron's response — committing to a $200 billion long-term capacity investment roadmap and ramping HBM4 production for 2027 delivery — is the correct strategic response to a structural shortage, but it cannot relieve near-term supply pressure. The result is a pricing environment where HBM commands premium margins that are dramatically improving the economics of the memory business, transforming Micron from a cyclical commodity semiconductor company into a structurally advantaged AI infrastructure provider — and justifying the fundamental rethinking of its valuation that UBS's upgrade represents.
The Competitive Landscape: Micron, SK Hynix and Samsung in the HBM Race
The HBM market is, at present, a three-supplier market — Micron, SK Hynix, and Samsung — with SK Hynix currently holding the dominant position by virtue of being the earliest and most aggressive investor in HBM3E production capacity. SK Hynix supplies the majority of Nvidia's current HBM requirements and has maintained a technology lead in HBM that it has used to secure long-term supply agreements with the most demanding buyers. Samsung, which supplies HBM to a broader customer base but has faced quality yield challenges with its HBM3E products, is working through qualification issues with Nvidia that have prevented it from capturing the market share its production scale would otherwise support. Micron has been the fastest-growing HBM supplier of the three, achieving Nvidia qualification for its HBM3E products and ramping production aggressively — but from a lower base than SK Hynix. The competitive dynamic among the three suppliers will determine how the enormous revenue and margin opportunity of the AI HBM market is distributed over the next three to five years, and Micron's $1 trillion milestone reflects the market's judgment that it has secured a durable position in a market that is structurally undersupplied for the foreseeable future.
The global HBM market is projected to grow from around $35 billion in 2025 to approximately $100 billion by 2028, according to analyst estimates that are, if anything, likely to be revised upward as AI infrastructure investment continues to accelerate. At those market size projections, Micron's share of a three-supplier market would generate revenue from HBM alone that approaches its entire current annual revenue from all memory products — a revenue mix transformation that justifies the fundamental rethinking of Micron's earnings power that UBS and increasingly the broader sell-side community are now modelling. The memory chip downturn of 2023 to 2024 — when Micron posted operating losses for multiple quarters — is receding from the narrative with remarkable speed. The structural case that HBM-driven AI demand has permanently changed the economics of the memory business is becoming the dominant investment framework, and Micron's trillion-dollar crossing is the market's most visible validation of that thesis.