# DC Hub — Q3 2026 Quarterly Deep-Dive
_Live data, 90-day window._ [Full report](https://dchub.cloud/reports/quarterly-deep) · [JSON](https://dchub.cloud/api/v1/reports/quarterly-deep) · CC-BY-4.0

## Executive summary
_auto-generated · claude-haiku-4-5-20251001 · 2026-07-10_

The defining shift in Q3 2026 was the decisive retreat from European markets coupled with a structural capital reallocation toward North American power-constrained regions. Dublin, London, Amsterdam, Frankfurt, and Manchester entered hard-avoid territory—five of Europe's historically prime colocation hubs now flagged as non-starters. Simultaneously, Williston (MISO), Cheyenne (WECC), Rural SPP, and Montréal emerged as the only six markets globally scoring BUILD verdicts, signaling an unprecedented geographic concentration of capital appetite. This wasn't gradual: 271 of 316 scored markets landed in AVOID, with only 39 in CAUTION. The verdict distribution alone—a 86% hard-avoid rate—reflects the market's collective recognition that power availability, not latency or incumbent density, now gates deployment. CoreWeave's $48 billion in sequential tranches during this window underscores that non-hyperscaler players are willing to lead where traditional cloud operators remain cautious on overbuilt EMEA infrastructure.

The window exposes a widening structural gap between hyperscaler risk tolerance and private-capital optionality. Hyperscaler billion-plus deals remain opaque in this period—the pipeline shows 340 announced facilities (14.3 GW) and 154 under construction (33.7 GW) but no published mega-acquisitions—suggesting internal build strategies dominate over external M&A for the largest players. Meanwhile, the M&A volume hit $1.98 trillion across 940 deals, with CoreWeave's aggressive posture capturing outsized mindshare. This reveals a two-tier market: AWS, Microsoft, and Google are vertically integrating power-first strategies with direct land and renewable contracts, while PE-backed and pure-play infrastructure funds are forced into either niche geographic arbitrage (Williston, Cheyenne) or emerging workload classes (AI inference, edge GPU clusters). Over the next two to four quarters, expect hyperscaler projects to drive capacity growth in power-abundant zones (SPP tier-2 nodes, Canadian Quebec), while PE capital fragments into smaller bolt-ons and EMEA secondary markets.

Monitor Cheyenne's execution velocity and Williston's PPA pipeline closely. Both markets have institutional interest but face grid interconnection timelines that could slip; early commissioning wins here will vindicate the BUILD thesis and trigger follow-on capital. Second, watch for CoreWeave's target deployment ratio and any competing mega-commitments from xAI, Lambda, or Crusoe—a sustained pattern of $40+ billion tranches to non-hyperscalers signals permanent tier-two infrastructure, not a funding anomaly. Third, track any reversion signals in EMEA: a single major UK or Dutch greenfield announcement with long-term power terms would suggest the AVOID verdicts are cyclical and tied to 2025–26 grid stress, not structural overcapacity. If that doesn't materialize by Q1 2027, European M&A will compress further and PE capital will fully pivot to North American power arbitrage.

## Headline numbers
- **21,937** facilities tracked
- **316** markets scored by DCPI
- **940** M&A deals ($1,985.3B disclosed) in last 90 days
- **114** press releases tracked

## Top BUILD markets
- **Williston, ND** (MISO) · DCPI —
- **Cheyenne, WY** (WECC) · DCPI —
- **Cheyenne** (WECC) · DCPI —
- **Rural SPP** (SPP) · DCPI —
- **Montréal** (HQ) · DCPI —

## Top AVOID flags
- **Dublin** · —
- **London** · —
- **Amsterdam** · —
- **Frankfurt** · —
- **Manchester** · —

## Hyperscaler $1B+ deals
- **?** acquired **?** · undisclosed
- **?** acquired **?** · undisclosed
- **?** acquired **?** · undisclosed
- **?** acquired **?** · undisclosed

## Attribution
DC Hub. (2026). Quarterly Deep Report. https://dchub.cloud/reports/quarterly-deep. Licensed CC-BY-4.0.

---
_Generated 2026-07-10 16:28:54 UTC · [/api/v1/reports/quarterly-deep](https://dchub.cloud/api/v1/reports/quarterly-deep) · [/llms.txt](https://dchub.cloud/llms.txt)_