Capital is finally chasing utilization, not logos, as specialized AI clouds turn GPU scarcity into strategy and push buyers to place each workload where it truly runs best, and Verda’s $117 million raise offers a timely lens on how training, advanced inference, and agentic patterns are rewriting
Lead A single 400G stream hits a switch, fans into APIs and storage calls, and suddenly the fabric groans under multi-terabit east-west traffic that no steady-state forecast ever predicted and no tidy oversubscription ratio can fully contain. That is the daily reality of AI training and data prep
Lead: The Rising Tide Meets the Server Rack Beneath flight paths and container cranes, where fiber landings lace the seabed and grid nodes hug the coast, a new class of data center is edging into view—floating, seawater-cooled, and positioned to feed AI’s hunger without squeezing the last buildable
Matilda Bailey has spent years at the sharp edge of networking and next‑gen wireless, where bandwidth, latency, and memory footprints decide what ships and what stalls. Today she unpacks Google Research’s TurboQuant, a claimed 6x reduction in inference memory and an 8x speedup on the same GPUs,
Permitting clocks increasingly decide which data centers power AI growth and which sink into multiyear limbo despite flawless engineering, because approvals now gate billions in capital, construction mobilization, and utility planning even when equipment lead times and grid interconnection appear
The center of gravity in AI is sliding from raw model speed to system choreography, where the slowest hop, not the fastest GPU, dictates tempo and turns infrastructure planning into a game of coordination rather than pure compute. That pivot matters because long-lived, tool-using agents have pushed
