Eight months. That is how long it took for the dominant AI narrative to swing from one extreme to the other.
January 2025: DeepSeek's new reasoning model triggered a tech selloff. Nvidia dropped nearly 17% in a single day, and the narrative shifted to "maybe we will not need that massive chip, datacenter, and power buildout after all." The logic was simple: if models could be that cheap and efficient, capex and power requirements might be far lower than expected.
October 2025: Fermi goes public with zero revenue, raising $682.5M at an initial valuation around $12.5–15B based on a plan to eventually supply up to 11 GW of power for AI datacenters (roughly five Hoover Dams if fully built). The story? Non-binding letters of intent, conditional land leases, and no tenant revenue expected until 2027. Shares surged on debut.
If January's DeepSeek moment whispered "descale," October's Fermi moment shouted "hyperscale." Same year, opposite conclusions: the hype cycle in fast forward.
Our view. AI is real, the money is real, and the capabilities are powerful. We use it every day. But adoption patterns, architecture choices, and who wins versus who loses remain uncertain. The right response is not to panic in January or throw out discipline in October, it is to take a long-term, data-driven approach that is responsive to new information, measured in position sizing, and grounded in a repeatable process.
Ignore narrative whiplash. Both January's "AI needs no power" and October's "AI needs all the power" can be partially wrong. Base your positions on evidence, not sentiment.
This is not the first market narrative to swing wildly. Having managed MLPs through their own boom-and-bust cycles, I have seen this movie before. The pattern repeats: early skepticism, euphoric adoption, overbuilding, shakeout, then a smaller set of durable winners. The goal is not to predict every twist. It is to end up owning more winners than losers by staying disciplined when the crowd swings from fear to frenzy.
At Cap Six, we believe long-term outperformance comes not from chasing narratives, but from a process built to adapt as facts change, without overreacting. Our investment framework is designed to stay grounded when market sentiment swings from one extreme to another.
We apply a consistent, data-driven approach that:
- Focuses on evidence, not emotion—evaluating thousands of data points to identify durable signals
- Maintains risk-aware portfolio construction—to avoid overexposure when markets overreach
- Continuously recalibrates—as new information surfaces, but without abandoning core discipline
Whether it is the hype around AI, energy, or any other macro theme, our role is to separate signal from noise and position client portfolios for sustainable, risk-adjusted returns, not headlines.