Buying the AI Bottlenecks: AMD, Micron and GE Vernova on a Market Pullback

Market Analysis

AI spending by hyperscalers and large enterprises remains the dominant force shaping semiconductor and energy markets. Wall Street estimates for AI capital expenditure have been rising; some analysts now model a 2026 AI capex opportunity near $800 billion. Big cloud providers are reflecting that reality in their budgets. Alphabet disclosed a $180190 billion AI infrastructure plan for calendar 2026 and is arranging roughly $80 billion of financing to support it. Meta flagged a higher capex outlook near $145 billion with memory pricing cited as a driver, and Microsoft raised capex to about $190 billion, saying roughly $25 billion of the increase stems from higher component costs.

That surge in spending is producing concentrated supply pressures across three inputs that matter most to AI deployments: compute (CPUs and server-grade GPUs/custom accelerators), memory (DRAM/HBM), and power/energy capacity for data centers. These constraints have created what I call AI bottleneck winnerscompanies that supply scarce yet essential inputs. The recent market pullback, after a strong run in AI-related equities, creates an opportunity to add exposure to a small portfolio of those bottleneck plays.

Investment Thesis and Top Picks

I am buying three names I view as structural beneficiaries of sustained AI capex: Advanced Micro Devices (AMD), Micron Technology (MU), and GE Vernova (GEV). Each addresses a distinct bottleneckserver CPUs and GPUs, high-bandwidth memory, and energy infrastructurewhile posting accelerating top-line dynamics that justify investor attention even after substantial rallies.

AMD. AMDs EPYC server CPU business has become the companys primary driver of data-center revenue growth. Management has materially revised its outlook: the company now expects its server CPU business to grow at roughly a 35% compound annual growth rate (up from an ~18% growth outlook a year earlier). AMD is also supplying server GPUs and has deals to deliver roughly 1 GW apiece to major customers such as Meta and OpenAI, which yields significant recognized revenue on deliveries. Street estimates show strong revenue acceleration into calendar 2026 (consensus growth near 43% for CY26), making AMDs forward multipleabout 15x CY26 saleslook rich on the surface but more reasonable given the rapid top-line acceleration. I maintain a Strong Buy view and am adding on dips.

Micron. Memory is a clear bottleneck for AI infrastructure. DRAM demand has surged, lifting DRAM and HBM pricing sharply over the last 912 months. Micron, one of the three largest DRAM manufacturers (alongside Samsung and SK Hynix), sits at the center of that supply tightness because DRAM is a core input for HBM used in GPUs and other accelerators. Industry reports indicate DRAM prices could rise another ~60% in Q2, reinforcing a favorable pricing backdrop. Micron was recently authorized, along with its peers, to supply HBM4 to Nvidias upcoming GPU platformsanother positive for memory demand. I remain bullish on Micron and am using pullbacks to add exposure.

GE Vernova. Power and energy capacity are less discussed in retail AI narratives but are becoming a binding constraint as hyperscalers add racks of power-hungry accelerators. GE Vernova is a major industrial energy franchise that serves data-center and utility-scale renewable needs; it is also one of the worlds few large wind-turbine manufacturers alongside Vestas and Siemens. GEVs revenue grew 9% to $38.1 billion in calendar 2025, and consensus midpoints for CY26 project roughly 19.5% revenue growth to about $45.5 billion. Backlog expanded rapidly: Q1 CY26 backlog was reported up about 62% to $120 billion. The company plans to add around 20 gigawatts of annualized output capacity by Q3, underscoring the scale of its addressable opportunity. At an approximate forward sales multiple near 5.4x, GEV looks attractively priced against accelerating fundamentals. I do not yet own GEV but plan to initiate a position on the current market pullback.

Risks & Considerations

All three names are highly levered to continued AI capex. If hyperscalers materially trim AI budgets or enterprise AI demand falters, component and equipment orders could be delayed or reduced, and the earnings outlook for these companies would deteriorate. Inflationary pressures, energy cost volatility, or geopolitical disruptions that affect supply chains (for example, commodity inputs or trade restrictions) could also change the demand-supply balance and weaken these stories.

My base case assumes the current pullback is transitory and that AI spending continues to expand. If instead the market enters a deeper correction driven by a structural reversal in AI investment, I will reassess positions and risk exposure.

Takeaway

The AI era is generating concentrated, measurable demand for CPUs, memory, and energy infrastructure. AMD, Micron, and GE Vernova represent three clear plays on those bottlenecks: server compute, DRAM/HBM supply, and power capacity/renewables respectively. Given the strength of hyperscaler budgets and rising market backlogs and pricing in these segments, I am bullish on all three and plan to use the current sell-off to increase exposureadding to AMD and Micron where I am already long, and initiating a position in GE Vernova.

Analyst disclosure: I have a beneficial long position in AMD and MU. I intend to open a position in GEV. This article reflects my views and is not personalized investment advice; investors should consider their own objectives and risks before acting.