Iren's Sweetwater Announcement Hints at a Larger Gigawatt Pipeline Than the Public Number

Iren's 2GW Sweetwater AI data-center hub was prioritized for ERCOT power allocation, while management reiterated delivery of its first 300MW in Q4 2027. At Iren's stated $20M-$25M annual revenue per MW, that initial capacity could support $6B-$7.5B of annual recurring revenue. The release also referenced additional large-scale projects not yet publicly disclosed, fueling speculation that Iren's development pipeline could materially exceed its disclosed 5.8GW, though timing, grid interconnection and funding remain key execution risks.

Analysis

The market is likely to capitalize IREN on a “powered-land” optionality framework before cash flow exists, but the critical valuation bridge is not headline gigawatts: it is contracted, deliverable MW multiplied by tenant credit quality and the cost of financing GPU and electrical infrastructure. Grid-priority status reduces one permitting bottleneck, yet does not establish transmission completion, interconnection cost, equipment procurement, customer prepayments, or acceptable project-level returns. This creates a 12-24 month gap in which equity upside is driven by contract announcements and downside by schedule or capex revisions.

The most important second-order effect is that scarce ERCOT power positions should widen the valuation dispersion among neocloud operators. IREN may deserve a premium to HUT and NUAI only if it converts capacity into signed AI-cloud contracts without materially dilutive equity issuance; otherwise, its opacity should warrant a discount rather than an option premium. The implied revenue-per-MW framing is especially aggressive absent disclosed utilization, GPU ownership versus customer-supplied hardware, power pass-through terms, and lease duration; investors should not annualize it as recurring revenue until these economics are independently disclosed.

Near term, the likely catalyst path is incremental site disclosure, interconnection agreements, and customer financing commitments rather than operational results. Over 6-18 months, the binding constraint may shift from power to capital: a buildout at this scale can turn a seemingly asset-light AI narrative into a leveraged infrastructure story just as GPU economics and hyperscaler demand normalize. Consensus appears to underweight this financing-duration mismatch while overvaluing undisclosed capacity as if it were already contracted.

Key Decisions for Investors

• Maintain IREN as a catalyst watch rather than add on capacity speculation. Upgrade to a 6-12 month long only after disclosure of binding tenant contracts, delivered-power dates, capex per MW, and financing terms; target at least 2x contracted-revenue coverage of incremental fixed charges as the underwriting threshold.
• If IREN rallies materially on further undisclosed-pipeline commentary without a customer contract, consider a 1-3 month relative-value short IREN versus long NBIS or NVDA. The thesis is that IREN’s multiple is more exposed to interconnection and funding de-risking, while NBIS/NVDA have more immediate demand monetization; cover on a binding IREN lease/prepayment announcement or definitive construction financing.
• Use HUT as a monitoring hedge rather than a clean long: a widening IREN/HUT valuation spread is tradable only after normalizing for energized capacity, debt, and hosting versus self-mining exposure. Enter long HUT/short IREN only if IREN’s enterprise value per contracted MW exceeds HUT’s by more than 50% without superior customer commitments.
• Set alerts for IREN guidance changes on first-power timing, capital raises, customer deposits, and power-cost pass-through. Any delay to initial delivery, equity issuance before tenant prepayments, or disclosed utilization below plan would falsify the premium-powered-land thesis and could drive rapid multiple compression.