That is not an engineering story. It is a commercial signal.
Vertiv's stock dropped the day AWS announced the IRHX. That tells you everything about what this market understands: when a hyperscaler stops buying from a supplier and starts building in-house, the upstream vendor loses revenue and the hyperscaler gains a cost structure advantage that compounds over every future deployment cycle.
Here is what Enki's signal data shows across 2023 to 2025:
In 2023, AWS was partnering to adopt liquid cooling,
Nvidia collaboration, reclaimed wastewater programs, third-party technology.
In 2024, AWS quietly acquired a nuclear-powered data center campus for $650M. No fanfare. A structural move.
In Q3 2025, they launched IRHX,
proprietary, rack-level, designed specifically for Nvidia Blackwell GPUs. PR volume hit its highest point of the two-year period. But the commercial event count was 2. The announcement did the work. The infrastructure was already in place.
This is the pattern that separates PR from execution: the capital decisions happen quietly, the announcement happens publicly, and by the time the market reacts, the strategic position is already locked in.
What to track now:
→ Whether competitors respond with equivalent proprietary cooling or stay dependent on Vertiv and Schneider
→ Whether IRHX becomes a licensing vehicle or stays internal
→ Whether the $20B nuclear initiative follows the same quiet-capital, loud-announcement pattern
The signal is not the announcement. The signal is what was already built before the announcement was made.