DLR, EQIX: Data Center REITs as the Silent AI Capex Beneficiary

Hyperscalers committed to $400-500 billion of US AI infrastructure capex through 2026. That capital flows through three layers — semiconductors (NVDA, AMD, AVGO), power generation (NRG, VST, CEG), and the colocation real estate that actually houses the compute. The third layer — data center REITs — has been the quietest large-cap beneficiary in the public-markets cohort. DLR closed June 3 at $183.50 with a $63 billion market cap, EQIX at $1,077 with $106 billion. Both names trade with operating fundamentals that suggest the AI capex thesis is materially underpriced versus the noise-heavier ASIC peers — especially after AVGO's June 3 sell-off showed how cohort multiples react to even modest guidance disappointment.

Why colocation is the silent leverage point

The data center REIT business model is the inverse of what hyperscaler customers do directly: instead of capex-heavy build-and-burn cycles, REITs build powered shells under long-dated leases (typically 10-15 years) and recognize stable revenue with predictable cap-ex amortization. When AI workloads scale, the marginal demand for "AI-ready" colocation space (200kW+ rack density, liquid cooling support, robust power redundancy) accelerates dramatically. DLR and EQIX have spent the past 18 months retrofitting and adding AI-ready capacity to meet this demand.

The financial signal: DLR's TTM EBITDA margin is 61.2%; EQIX is 45.5%. Both elevated. DLR's free cash flow margin (TTM) is 18.0%; EQIX is 9.4%. EQIX's lower FCF reflects ongoing growth capex — most of which is currently being deployed in AI-ready expansion (the FY25 FCF was negative $400M, indicating they are spending ahead of revenue). DLR's higher FCF reflects a more measured pace of expansion. Both companies pay dividends — DLR yields 2.66%, EQIX yields 1.83%.

The valuation gap with the AI infrastructure peer cohort is wide. AVGO trades at 18.4× forward P/S; NVDA at 19.2×; MRVL at 59.5×. DLR at 9.25× forward P/S and EQIX at 10.2× sit roughly half the forward multiple of the ASIC names — and pay dividends those ASIC names don't.

Data points

drillr terminal snapshot (June 3, 2026):

A specific tape divergence is worth flagging. EQIX outperformed DLR by 19 percentage points YTD and by 12 percentage points over twelve months. The performance gap reflects EQIX's more aggressive AI-ready capacity expansion strategy: EQIX has announced $5+ billion of AI-specific facility investments through 2027, while DLR's announcements have been more measured. The market has priced EQIX's higher growth trajectory but has not yet priced DLR's defensive yield characteristics. The dividend yield differential (2.66% vs 1.83%) reflects this.

EQIX's revenue mix is critical: digital interconnection (cross-connect, peering exchange, and direct cloud connections) represents approximately 18% of revenue but generates 60%+ gross margin. As AI workloads increasingly demand low-latency multi-cloud connectivity, this segment's growth rate accelerates well above the colocation base — pulling overall EBITDA margins higher.

{
  "hint": "A clean infographic showing AI capex flow as a horizontal funnel. Top label: 'Hyperscalers MSFT GOOGL AMZN META 2026 capex $400-500B'. The funnel narrows downward through three labeled rectangular layers: 'Semiconductors NVDA AVGO AMD' (top, blue), 'Power Generation NRG VST CEG' (middle, gold), 'Colocation REITs DLR EQIX' (bottom, green). Each layer has a small dollar callout for its addressable revenue. Light gray background, no decorative elements, business-publication aesthetic, minimal text.",
  "aspect": "9:16",
  "style": "minimalist infographic, clean financial-publication style",
  "alt": "AI infrastructure capex flow showing how hyperscaler dollars cascade through semiconductors, power generation, and colocation REIT layers including DLR and EQIX",
  "caption": "AI capex flow — three layers that absorb hyperscaler spending"
}

Analysis: positioning the cohort

Three scenarios for the DLR / EQIX trade over 6-12 months.

Scenario A — AI infrastructure capex sustains through 2027. The base case implied by current consensus. Hyperscalers continue spending; AI-ready colocation demand outstrips supply; both DLR and EQIX deliver upside revisions to FY26 forward revenue growth (+6% and +10% currently). DLR fair value lands $205-$215; EQIX $1,150-$1,200. Implied upside: 12-17% over 12 months plus dividends.

Scenario B — Hyperscaler capex moderates by H2 2026. The Cramer-flagged "capital supply" concern materializes; one or more hyperscalers (most likely AMZN given its cultural rift around capex) slows AI infrastructure spending growth. Lease-renewal pricing power weakens. DLR and EQIX hold at current levels — defensive yield characteristics preserve downside. Total return: 2-4% plus dividends.

Scenario C — Power infrastructure becomes the binding constraint. AI workloads consume so much electricity that data center expansion can't keep pace with hyperscaler demand; the colocation REIT cohort raises rents materially. DLR and EQIX accelerate revenue growth beyond consensus; forward multiples expand toward 13-14× P/S. Implied upside: 25-35%. This scenario depends partly on whether AI power demand stocks successfully deliver new megawatts on schedule.

The risk-reward is asymmetric on the upside. Scenario A is the modal case; Scenario B has modest downside (defended by dividends and below-cohort multiples); Scenario C is the asymmetric kicker. DLR has the better risk-reward for yield-focused allocations; EQIX has the better risk-reward for growth-focused allocations.

What to watch

• EQIX Q2 2026 earnings (early August): Watch interconnection-revenue growth and AI-ready capacity utilization. Numbers >25% YoY for the interconnection line confirm Scenario A; below 20% suggests Scenario B.
• DLR FY26 H1 forward lease commitments: DLR typically discloses new leases and customer mix. Watch for hyperscaler-named leases (without disclosing customer identity) — a step-change in committed capacity signals demand acceleration.
• Hyperscaler quarterly capex prints: MSFT (mid-July), GOOGL (late July), AMZN (early August), META (late July) all guide capex. Aggregate growth above +12% YoY sustains the AI-infrastructure thesis through 2027.
• AI-ready power capacity additions: data center electricity demand growth at 7-9% per year (well above the 1-2% historical baseline) means new generation must come online to support DLR/EQIX expansion plans.
• Rate path and REIT cap-rate dynamics: data center REITs trade partly on interest rates. A meaningful steepening at the long end (10-year Treasury above 4.75%) would pressure REIT multiples generally, though the AI-specific growth thesis can offset.

---