Everyone is watching the chip stocks. Fair enough — the numbers there are extraordinary. But somewhere between the GPU rack and the utility bill, there’s an entire layer of the AI infrastructure build that is getting less attention and, arguably, offering a different risk/reward profile. Two names sit at the center of it: Vertiv (VRT) and Cameco (CCJ).
One supplies the power and cooling infrastructure inside the data center. The other supplies the uranium fuel for the nuclear reactors that are increasingly being contracted to power them. Different companies, different sectors, same underlying demand driver.
Vertiv: The Thermal Management Trade
Liquid cooling is no longer a niche. It is the default architecture for new AI data center builds. Traditional air cooling systems are physically insufficient for the power densities required by modern GPU clusters — a problem that becomes more acute with every generation of accelerator hardware. The global data center liquid cooling market is projected to grow from $5.7 billion in 2026 to $29.2 billion by 2033, according to Persistence Market Research.
Vertiv is the cleanest public proxy for this buildout. Q1 2026 net sales hit $2.65 billion — up 30% year over year. The Americas region grew 44% organically, driven by data-center demand. Adjusted diluted EPS grew 83% year over year. Net income reached $390 million on that revenue base. The project backlog stands above $15 billion — roughly 12 to 18 months of forward revenue at current run rates.
Management has been executing an acquisition strategy to deepen its competitive position. In March 2026, Vertiv agreed to acquire ThermoKey, a heat rejection and heat-exchange specialist. In April, it acquired Strategic Thermal Labs — a specialist in server-side liquid cooling — strengthening its engineering capability at the interface between the server rack and broader data center infrastructure. These aren’t defensive moves. They’re moat-widening moves ahead of a demand wave that is still in its early innings.
Full-year 2026 guidance implies approximately 34% sales growth. Management targets $13.25–$13.75 billion in revenue and $6.30–$6.40 adjusted diluted EPS for the year. The stock reports Q2 2026 results on Wednesday, July 29, 2026. It will get less attention. That’s often where interesting things happen.
The valuation is not cheap. VRT trades near $333, with a forward P/E around 48x. That multiple requires execution to continue — any guidance miss would be punished quickly. But on a PEG-adjusted basis, with EPS growing 43% and a backlog that provides unusual revenue visibility, the premium is defensible as long as hyperscaler capex stays elevated.
Cameco: The Uranium Fuel Supply Trade
Here’s where it gets interesting. The hyperscalers are not just contracting for power — they’re contracting for nuclear power specifically. Four AI hyperscalers collectively plan roughly $710 billion in 2026 capex, and nuclear is increasingly the preferred source of 24/7 carbon-free baseload electricity. Meta has announced agreements to help unlock up to 6.6 gigawatts tied to nuclear energy projects. Microsoft, Google, and Amazon have all signed long-term power purchase agreements with nuclear operators. Thirty-eight countries have pledged to triple nuclear capacity by 2050.
That demand runs directly through uranium. The long-term uranium price has climbed to approximately $91.50 per pound. Kazatomprom — the world’s largest uranium producer — intentionally cut 2026 production guidance by 10% to support pricing, functioning as a “nuclear OPEC.” All three major HBM chip producers have committed their capacity through 2026, and the uranium supply picture is similarly constrained.
Cameco (CCJ) is the world’s largest publicly traded uranium company and owns 49% of Westinghouse — the AP1000 reactor OEM. Q1 2026 results showed adjusted net earnings increased to $203 million, with Westinghouse adjusted EBITDA jumping 33% to $122 million. The uranium segment drove C$423 million in adjusted EBITDA versus C$286 million a year earlier. Average realized uranium price was $66.21 per pound, up from $62.55 in the prior year. By end of 2025, Cameco had secured long-term contracts to deliver approximately over 28 million pounds of U3O8 per year on average over the next five years, with commitments higher than the average in 2026 through 2028, and lower than the average in 2029 and 2030.
The US DOE has flagged up to $17.5 billion in nuclear supply chain loans via conditional commitments. The policy tailwind is as strong as it has been in decades. Some bull-case uranium price forecasts see $120 per pound by late 2026.
Technical Framework
VRT has been consolidating in the $310–$340 range since its June run. Support is near $315. A break above $345 with volume opens a path toward prior highs. The July 29 earnings report is the near-term catalyst. CCJ has pulled back from its 52-week high of $135 and recently traded near $101 — still up approximately 50% over the past year. The uranium ETF (URA) returned 120% over the past year. The pullback in CCJ represents a potential re-entry point for traders who missed the initial move.
Scenario Modeling
Bull Case (VRT): Q2 results show continued 30%+ revenue growth, backlog expands beyond $15 billion, and management raises FY2026 guidance. Market re-rates toward 55–60x forward earnings as AI capex visibility extends. Target: $380–$400.
Base Case (VRT): Q2 meets guidance, backlog stable, margins hold. Stock moves toward $350–$360 post-earnings. Continued hyperscaler capex commitments sustain demand narrative through 2026.
Bear Case (VRT): Any signal that hyperscaler cooling budgets are moderating, or that competitors (Eaton, Schneider Electric) are taking share in liquid cooling, sends the stock back toward $290–$300. High valuation means little forgiveness for misses.
Bull Case (CCJ): Uranium price moves toward $100–$120 on supply constraints and accelerating nuclear contracting. New hyperscaler nuclear PPAs announced. CCJ targets prior 52-week highs and beyond.
Bear Case (CCJ): Iran ceasefire holds, energy prices broadly decline, nuclear enthusiasm cools with sentiment. CCJ retraces toward $85–$90 range.
The Framework
The AI infrastructure trade has three layers: the chip, the cloud, and the physical plant. Most capital has flooded into the first two. The physical layer — power, cooling, and fuel — is where the demand is becoming inelastic. You can delay a GPU order. You can’t delay the electricity bill.
Both VRT and CCJ represent different entry points into the same underlying demand thesis. Together, they function as a hedge on the AI build continuing — not a bet on which model or which chip wins, but a bet that the build itself keeps happening. That’s a different kind of conviction, and right now, it has a lot of data behind it.
For informational and educational purposes only. Not investment advice. Trading involves risk, including loss of principal.
