Micron’s Open-Source AI Bet: Why Bank of America’s Top Pick Is Riding a Different Kind of Chip Boom

Micron Technology gains 12.64% as Bank of America highlights open-weight AI models driving massive memory demand, with HBM capacity sold out through 2027.

The memory-chip maker from Boise, Idaho, is rewriting the rules of the semiconductor playbook. Micron Technology isn’t just riding the AI wave—it’s betting that the future belongs to open-source models that multiply hardware demand in ways closed systems never could. After a 12.64% surge on Tuesday to close at €852.80, the stock is signaling that investors are starting to buy into that thesis.

Bank of America’s decision to add Micron to its “US 1” list of top-conviction picks provided the immediate catalyst. But the underlying logic runs deeper than a single analyst upgrade. The bank’s argument hinges on open-weight AI models—systems like Moonshot AI’s newly launched Kimi K3, which can be downloaded and run locally on any company’s own infrastructure. Unlike the walled gardens of hyperscaler APIs, each local installation of Kimi K3 demands its own dedicated memory. With 2.8 trillion parameters requiring roughly 1.4 terabytes of High Bandwidth Memory spread across at least 64 AI accelerators, the storage footprint of these models is staggering.

The Multiplier Effect of Open Models

Closed AI systems concentrate computing power in a handful of data centers. Open models do the opposite: every download, every local deployment creates fresh demand for HBM, DRAM, and NAND. A single model multiplies itself across thousands of hardware instances. That dynamic explains why Micron has surged 817.58% over the past twelve months—even after shedding 19.26% in the last 30 days.

The phenomenon has a name: “chipflation.” Samsung, SK Hynix, and Micron have shifted nearly all their production capacity toward HBM for AI data centers. The knock-on effect for the PC market has been brutal—memory prices for consumers and businesses have exploded 400% in a single year.

Capacity Sold Out Through 2027

Micron’s entire HBM output for 2026 is already spoken for, and large portions of 2027 are under contract. The company has locked in 16 multi-year customer agreements with price floors that extend through 2030—an extraordinary level of visibility in an industry notorious for boom-and-bust cycles. This planning security is rare; memory prices typically swing wildly, and customers usually hesitate to commit to long-term deals.

The competitive landscape reinforces Micron’s position. Chinese rival ChangXin Memory Technologies (CXMT) is expanding rapidly and posting strong quarterly profits, with market valuations reportedly reaching 2 to 3 trillion yuan. But CXMT remains confined to commodity DRAM. The lucrative HBM3E and HBM4 segments remain an effective three-player oligopoly shared by Samsung, SK Hynix, and Micron.

The Buyback Clock Is Ticking

For long-term investors, one date stands out: December 2026. That’s when the CHIPS Act restrictions on Micron’s share repurchases expire. Analysts at both Bank of America and UBS see the company entering a phase of massive free cash flow generation. Some estimates suggest Micron could buy back 5% to 6% of its market capitalization annually once the restraints lift—a structural demand driver for the stock, not merely a cosmetic signal.

At a current market cap of roughly €840 billion, such buyback capacity would be unprecedented for a semiconductor firm. The analyst consensus price target of €1,306.97 implies upside potential of 53.3%, though the stock remains 22.73% below its 52-week high of €1,103.80 reached in late June.

Volatility as a Feature, Not a Bug

The annualized volatility of 108.68% over the past year is a stark reminder that this is not a stock for the faint-hearted. The shares currently trade just 2.46% above their 50-day moving average of €832.45, suggesting a tentative stabilization after a period of intense profit-taking.

The coming weeks will bring quarterly results from major tech companies, offering a clearer picture of how much capital is actually flowing into AI infrastructure. For Micron, the near-term price direction matters less than the structural question: Is the demand surge from open AI models as durable as the current valuation implies? The 16 long-term contracts suggest the market is already voting yes.