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. 21,738 facilities · 1812.3 GW · 80 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.
Anthropic's $35 billion data-center acquisition on September 1st anchored a month that saw 80 deals worth $59.5 billion across 57,960 megawatts—a 9.6 percent increase in deal count month-over-month, though total value fell 15 percent as the hyperscaler bet outsized the rest of the market. Behind Anthropic's move, Equinix closed a $10 billion acquisition from KKR and Stack purchased GIC's portfolio for $5 billion, signaling aggressive consolidation among operators seeking scale in power-constrained markets. The global footprint reached 21,738 facilities totaling 1.81 terawatts, with new construction adding 2,410 facilities in the month alone. Richland Parish and a single unnamed facility each claimed 5,000 megawatts of capacity, while Ashburn—still the densest market—held 4,296 megawatts across 161 sites, and Las Cruces and Abilene combined for 7,450 megawatts of newer, lower-cost infrastructure in the Southwest corridor.
The spike in deal volume despite lower aggregate value points to a bifurcation: mega-deals by AI leaders securing long-term power access, and smaller tactical acquisitions by operators filling regional gaps. A 7,900 percent year-over-year surge in deal count—while partly a statistical artifact of a thin September 2025 base—reflects the structural shift toward consolidation as power scarcity tightens. The near-term signal is clear: capacity will concentrate among operators with direct power offtake or utility partnerships, and the 369 gigawatt pipeline will fill selectively. Expect M&A velocity to remain elevated through Q4 2026, with secondary-market operators either absorbed or forced into niche strategies around renewable integration or latency-critical workloads. Richland Parish and the Southwest's emerging footprint are now proving grounds for whether new supply can outpace AI demand.
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 entered a decisive reprieve from European saturation, with Q3 2026 marking the first sustained rotation of capital away from Dublin, London, and Amsterdam toward North American power-constrained regions. Of 333 markets assessed this quarter, only 24 earned BUILD verdicts—a stark filtering that reflects neither shrinking demand nor retreating hyperscalers, but rather a hardened discipline around interconnection timelines and grid capacity. Meta deployed $150 million across three separate acquisitions, while CoreWeave committed $96 million to secure AI-optimized capacity, demonstrating that dry powder remains abundant but destination selectivity has tightened. The verdict distribution—223 AVOID, 86 CAUTION, 24 BUILD—reads as a market consolidating around scarcity, not abundance. This is the sound of a sector that no longer assumes all land with fiber is created equal.
The structural story is a power gap masquerading as a real estate problem. Williston, North Dakota; Upper Peninsula Michigan; Midland–Odessa; Rural SPP; and Cheyenne now rank as the only markets worth greenfielding, and every single one was selected for grid headroom, not tenant density. This inverts the playbook that governed 2023–2025. The announced pipeline holds 15,829 MW across 8,065 facilities, yet only 27,896 MW sits in planned status—a ratio that signals demand far exceeds authorized supply. Hyperscalers are not negotiating down anymore; they are negotiating exclusivity with regional utilities and RTOs. PE-backed regional operators face a two-tier outcome: those with contracted power in ERCOT, MISO, or SPP will command 12–18-month wait lists; those without will struggle to attract anchor tenants. M&A velocity—896 deals worth $2.16 trillion in 90 days—reflects consolidation around operational assets with grid certainty, not greenfield optionality.
Monitor Midland–Odessa and Upper Peninsula Michigan for hyperscaler announcement density over the next two quarters. These markets will absorb the first wave of AI-driven capacity demand that Dublin cannot serve, and their buildout velocity will set precedent for which regional operators secure long-term hyperscaler commitments. Watch European market absorption rates in parallel: if Q4 2026 sees continued AVOID verdicts in London and Frankfurt despite announced supply, Europe's data center surplus has hardened into structural overcapacity, forcing PE sponsors to discount yields or pivot to infrastructure plays. Finally, track SPP utility filings—any grid operator green-lighting 2+ GW of new interconnection capacity will become the next Midland, immediately repricing adjacent real estate.
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