Quick note after reading up on the AI data-center power bottleneck (HV transformers 3–5 year lead times, BYOP trend, etc.).
Main names I’m tracking:
\- \*\*Bloom Energy (BE)\*\* – big Oracle deal, fuel cells avoid HV/MV transformers, cleaner/quieter, faster to deploy than turbines.
\- \*\*Capstone Green Energy (CGEH)\*\* – microturbines (65 kW–1 MW), new 800 VDC version with Microgrids 4 AI for AI racks. Feels more like an edge/smaller data-center play; lower efficiency, but interesting niche.
\- \*\*FuelCell Energy (FCEL)\*\* – megawatt-scale fuel cells, 4+ GW pipeline, talking directly about data centers. Higher risk financially, but similar “on-site baseload” story to Bloom.
These are all on my watchlist now. The thesis is simple: whoever can deliver clean, fast, on-site power to AI data centers wins.
For the community:
\- Anyone have experience or deeper views on \*\*CGEH vs FCEL vs BE\*\* (or other small caps in this space like POWL, MOD, NVT, etc.)?
\- Do you think the “on-site power for AI” theme is still early, or are we already late?
\- Any other small-cap names you’re watching that could benefit from this bottleneck?
Curious for opinions, red flags, or anything I’m missing.
Sources: Substack Tech Fund