Live data-center market intelligence in WSJ/Bloomberg lede voice. Quote any sentence with attribution to DC Hub (CC-BY-4.0 — no NDA, no embargo, no license fee). Daily refresh, 300+ markets, 13,000+ deals tracked. Pre-written quotables in /reports/monthly.md if you want to skip the curl. 21,456 facilities · 849.7 GW · 256 deals this month.
curl https://dchub.cloud/api/v1/dcpi/scores?verdict=BUILD&limit=5
Tuned for reporters: punchy ledes, named markets, no jargon. Each sentence is quote-ready.
Blackstone's $70 billion acquisition of Anthropic's data-center portfolio in a single week marked the largest hyperscaler infrastructure consolidation on record, signaling that financial sponsors have moved from sideline observers to primary architects of AI compute capacity. June saw 256 deals worth $708 billion across 358 gigawatts—a 1,064% month-over-month surge in transaction count and the clearest evidence yet that the era of captive, company-owned AI infrastructure is contracting in favor of third-party ownership models. Richland Parish, Louisiana and the unnamed market "One" each absorbed 5,000 MW of the month's volume, while Ashburn retained its position as the densest market with 4,296 MW across 161 facilities. The velocity of deal-making—up 12,700% year-over-year—reflects not speculative froth but structural repricing: hyperscalers are monetizing depreciated assets and outsourcing the capital intensity of power procurement to operators and funds with lower cost of capital.
This pivot foreshadows a two-quarter correction in self-built hyperscaler announcements and a corresponding surge in lease-backed expansion in constrained markets like Ashburn, where per-facility density suggests limited greenfield runway. The Blackstone deal alone represents roughly 42% of the month's transaction value, indicating that mega-funds are now bidding aggressively for stabilized, operating capacity rather than participating in development risk. Expect continued M&A consolidation among regional operators and a marked acceleration in power-purchase agreements tied to merchant data-center platforms—the financial engineering of compute is displacing the engineering, at least as a headline-driver. Queue depth and grid-operator interconnection timelines, not real-estate availability, will become the binding constraints on Q3 and Q4 capacity announcements.
curl -s https://dchub.cloud/api/v1/reports/monthly/narrative | jq .narrative
350-word structural read across 90 days of capital, capacity, and verdicts. The structural-shift section is where we go beyond CBRE's H2 outlook.
The data-center market has bifurcated sharply along geography and power access, with 172 of 306 markets rated as cautionary and 116 outright avoided, leaving just 18 buildable locations—all concentrated in the interior United States where power grids can absorb new demand. The verdict is unambiguous: the 90 days through mid-June exposed European markets (Dublin, London, Amsterdam, Frankfurt, Manchester) as capital traps, while Williston, North Dakota; Cheyenne, Wyoming; and rural SPP territory have become the only defensible landing zones for new capacity. This is not a cyclical preference. The operational base remains vast—11,063 facilities generating 91 gigawatts—but pipeline velocity has stalled: only 154 facilities are under construction (33.7 GW total), and just 91 planned sites contain 27.9 GW. The market is consolidating, not expanding.
Private-equity capital, notably Blackstone's three separate $35 million tranches and a $30 million follow-on in the last 90 days, is now chasing portfolio depth rather than greenfield deployment. Across 332 M&A transactions worth $762 billion, the signal is clear: scale matters more than location. AirTrunk's $30 million entry sits alongside Blackstone's dominance because existing operating assets—concentrated in the hands of AWS (14.1 GW across 504 facilities), Microsoft (11.1 GW, 301 facilities), and Equinix (7.1 GW, 659 facilities)—have become the real asset class. The hyperscaler capital commitments above $1 billion remain redacted or incomplete in public disclosure, but the absence of announced mega-deals in this window, coupled with 69 press releases that disclosed little material capacity, suggests that the trillion-dollar build phase is entering a financing bottleneck. Hyperscalers are pivoting to M&A and partnership rather than owned capacity.
Watch Williston, North Dakota, and the Upper Peninsula of Michigan closely over the next two quarters. These MISO markets represent the last scalable power corridors east of the Mississippi, and their build verdicts will determine whether new capacity clusters along hydro and wind resources or stalls entirely. Simultaneously, monitor Dublin and London for divestiture signals—European assets will trade at a discount as PE holders face pressure to reallocate capital westward. The real tell: if no announced facilities cross the $1 billion threshold in Q3 2026, the market has entered a structural consolidation phase, not a cyclical pause.
curl -s https://dchub.cloud/api/v1/reports/quarterly-deep/narrative | jq .narrative
Three integration patterns, ordered easiest → most-integrated:
curl -s https://dchub.cloud/api/v1/reports/monthly/narrative \ | jq -r '.narrative'
curl https://dchub.cloud/reports/monthly.md
Returns the full monthly report as paste-ready markdown. Drop directly into Substack, Ghost, Notion, or your CMS.
Endpoint: https://dchub.cloud/mcp Auth: X-API-Key: <your key from /pricing> Tools: 27 (search_facilities, get_market_dcpi_rank, compare_isos, ...) Spec: https://dchub.cloud/llms.txt
Claude.ai, Claude Code, and any MCP-aware agent can pull live data directly.
DC Hub. (2026). https://dchub.cloud. Licensed CC-BY-4.0.DC Hub · dchub.cloud · Pricing · /llms.txt · /mcp