53 megawatts. Nine markets. $1 billion committed.
I Squared Capital announced a definitive agreement to acquire 10 Cogent Fiber data center facilities from Cogent Communications for $225 million, per a Cogent Communications SEC 8-K filed May 26, 2026 – the T1 anchor source for this deal. The transaction is expected to close in Q3 2026, pending Hart-Scott-Rodino antitrust clearance. It hasn’t closed yet. That’s a material contingency in any infrastructure deal of this scale.
Per I Squared Capital’s announcement, the acquired portfolio spans 9 U.S. markets, Chicago, Atlanta, Phoenix, Los Angeles, Kansas City, Baltimore/Elkridge, Houston, Nashville, and the Anaheim/Burbank/Stockton corridor, with 53 megawatts of installed power capacity and 259,000 square feet of colocation space. The company announced a commitment of up to $1 billion to develop the platform through infrastructure upgrades, customer expansions, and additional acquisitions.
The inference edge framing is the investment thesis. Analysts have interpreted this deal as reflecting a strategic shift toward distributed edge inference infrastructure, compute capacity positioned close to end users for low-latency AI workloads, as distinct from centralized hyperscale training centers. That’s analyst interpretation, not a stated I Squared strategic priority. But the portfolio geography makes the framing credible: these aren’t hyperscale campuses. They’re mid-market colocation facilities in major metro corridors, which is exactly the footprint inference-edge infrastructure requires.
The real story is the structure. A $225 million seed acquisition backed by a $1 billion platform commitment is a bet that edge inference infrastructure is undersupplied relative to incoming demand – and that assembling a distributed portfolio now, before that demand materializes, is the entry point. That’s a different investment logic than hyperscale data center development, which concentrates capital in massive facilities. This is the distributed version of the same infrastructure thesis.
This is the second infrastructure-anchored AI capital deal announced May 26, alongside Stord’s $250 million Series F targeting physical AI in logistics. The pattern from the hub’s April capital flow tracking holds: infrastructure and physical AI are absorbing a growing share of late-stage investment, as software-layer AI valuations face more scrutiny. The hub’s prior coverage of AI inference cost dynamics provides context for why distributed edge facilities are the next infrastructure build-out target.
What to Watch
What to watch
HSR clearance timeline for Q3 2026 close. Post-close, watch for I Squared’s announcement of specific infrastructure upgrade commitments and whether additional Cogent facility acquisitions follow, the $1 billion platform commitment implies this is a seed position in a larger portfolio-building strategy.
TJS synthesis
I Squared’s bet is a directional read on where AI inference demand is going. Centralized hyperscale compute serves training and large-scale inference workloads. Distributed edge colocation, 53 MW across 9 markets, serves latency-sensitive inference at the application layer. If enterprise AI adoption continues toward real-time, user-facing applications (agents, copilots, real-time analytics), the demand for low-latency compute close to users will follow. I Squared is positioning for that demand curve before it peaks. Watch the Q3 2026 close and the first post-acquisition upgrade announcement for confirmation that the $1 billion commitment is moving from announcement to deployment.