Stifel Conference | January 22, 2026 | Audience Q&A Prep
Fireside Chat with Blue Owl
Perspectives From a Leading Data Center Investor & Operator
Core Economics
$10-12B
Revenue / GW / Year
6 mo
= $1B+ Opportunity Cost
40-64%
Overbuild for N+1+1
2029
GE/Siemens Booked To
Blue Owl Real Assets
AUM
$75B
Capital Deployed
$100B+
Transactions
50+
Typical Lease Term
15-20 yrs
Dev Platforms
Stack, EEL
Market Dynamics (ERCOT)
Load Requests / Month
Tens of GW
Approved (Last 12mo)
~1 GW
US Total Requests
~1 TW
US Peak Load
750 GW
Nvidia 2024 Shipments
5 GW
Blue Owl Underwriting Framework
1. Credit
Investment-grade counterparties. MSFT (AAA), GOOGL, AMZN, ORCL. "If we're wrong on credit, nothing else matters."
2. Mission Criticality
Revenue-generating assets, not just cost centers. Contracted workloads before shovel hits ground.
3. Real Estate
Core markets (NoVA primary), great dirt, power capacity under control. Residual value protection.
The xAI Playbook
- PIONEER 100K GPU cluster in 4 months - bypassed grid entirely
- Bridge Power: Rent turbines from Solaris, VoltaGrid truck-mounted engines
- Deployed 500MW+ of turbines near Memphis
- Site selection: Built at TN/MS border - if one state blocks permit, pivot to other
- Paired turbines with Tesla Megapacks for load smoothing
"Speed is the moat. Getting 400MW online 6 months earlier is worth billions."
Following the Playbook
- ORCL/OPENAI 2.3 GW order - largest ever for onsite gas
- META Ohio deployment with Williams - 5 different turbine/engine types ("deploy whatever arrives")
- DOOSAN 1.9 GW order for xAI - Korean H-class entrant
- BOOM 1.2 GW with Crusoe - supersonic jet company pivoting to turbines
- WÄRTSILÄ 800MW US datacenter contracts - ship engine manufacturer
Equipment Landscape: Capex & Lead Times
| Technology |
$/kW |
Lead Time |
Ramp Rate |
Efficiency |
Notes |
| Aeroderivatives (GE LM2500/6000) |
$1,700-2,000 |
18-36 mo |
5-10 min |
~40% |
Can be truck-mounted, fast install |
| IGTs (Solar Titan, Siemens SGT-800) |
$1,500-1,800 |
12-36 mo |
~20 min |
~38% |
Refurb can cut lead time to <12mo |
| RICE - Medium Speed (Wärtsilä, Bergen) |
$1,700-2,000 |
15-24 mo |
~10 min |
~45% |
Same engines as cruise ships |
| RICE - High Speed (Jenbacher) |
$1,700-2,000 |
12-18 mo |
~10 min |
~42% |
VoltaGrid's preferred; containerized |
| Bloom Fuel Cells (SOFC) |
$3,000-4,000 |
Weeks |
Slow |
~55% |
No air permitting; 5-6yr cell life |
| H-Class CCGT (GE HA, Siemens HL) |
$2,200-2,800 |
24-36+ mo |
30-60 min |
>60% |
Best efficiency; longest install |
Redundancy Economics
- N+1+1 Overbuild: Required for 99.9% uptime off-grid
- VoltaGrid/Vantage: 1.4GW DC powered by 2.3GW generation (64% overbuild)
- PUE Factor: 1.4x for cooling overhead
- Redundancy Factor: Additional 10-17% for maintenance cycles
- Workarounds: Grid-as-backup, batteries, mixed turbine fleets
Supply Chain Bottlenecks
- Turbine Blades: Monocrystalline nickel alloys w/ rhenium, cobalt, yttrium
- Blade Suppliers: PCC, Howmet, CPP, Doncasters (fraction of customer size)
- Yttrium: Under Chinese export controls
- Heavy-duty logistics: 300-500 ton cores need specialized barges/rail
- Manufacturer PTSD: 30 years of boom-bust cycles = cautious expansion
QUESTION 01 | BRIDGE POWER & REAL ESTATE PILLAR
"The xAI Memphis playbook is spreading - AI labs bypassing the grid with rental turbines and truck-mounted engines, converting to backup once grid arrives. If tenants increasingly bring their own generation, does that structurally change your 'real estate pillar' calculus? What's the residual value of your land position when power is tenant-supplied?"
Why it matters: Blue Owl's three-pillar framework assumes power access is part of land value. If BYOG becomes standard, the moat shifts. Tests whether they're pricing in this structural change or still underwriting as if grid access = durable advantage.
QUESTION 02 | EQUIPMENT BOTTLENECK & SUPPLIER RELATIONSHIPS
"GE Vernova and Siemens are booked through 2029. New entrants like Doosan and Boom Supersonic are filling the gap. Do your development platforms - Stack and EEL - have preferential supplier relationships or reserved slots? Is equipment procurement becoming as important as land and power in your deal pipeline?"
Why it matters: Equipment lead times now rival interconnection timelines as the binding constraint. Probes whether Blue Owl's vertical integration extends to generation equipment or if they're exposed to the same queues as everyone else.
QUESTION 03 | CRYPTO CONVERSION & BROWNFIELD OPPORTUNITY
"Crypto miners hold existing power interconnections, ERCOT relationships, and staff who understand grid topology. SemiAnalysis notes they're 'really well positioned' for conversion. What inning are we in on retrofitting crypto sites for AI? Does the 64% overbuild requirement for off-grid redundancy make brownfield crypto conversion more attractive than greenfield?"
Why it matters: Crypto→AI is a near-term capacity unlock that's underappreciated by traditional DC investors. Probes whether Blue Owl sees this as viable pipeline or views crypto assets as lower quality / wrong locations.
Blue Owl Key Quotes
"There is nothing or nearly nothing happening on spec. Every asset we build has a long-term offtake with an investment-grade counterparty."
"Oracle's cloud contracts over the last 18 months are expected to grow their revenue by 30% over the next decade."
"The biggest limiting factor is not power. It's talent. There's simply not enough skilled labor."
"We are myopically focused on partnering with hyperscalers. CoreWeave is probably the one neocloud having real advantage without additional credit support."
SemiAnalysis Key Quotes
"AI cloud revenue can net $10-12M per MW annually. Getting 200MW online six months earlier can net $1-1.2 billion in revenue."
"Elon didn't even buy turbines - he rented from Solaris to bypass lead time. He also leveraged VoltaGrid's fleet of mobile truck-mounted engines."
"Meta's Ohio deployment with Williams comprises five different types of turbines & engines - clearly 'deploy whatever I can get on time!'"
"Crypto miners have staff whose energy person used to work for ERCOT. They know the underlying grid infrastructure."
Comparative Framework: 2000 vs. Today
Dot-Com Era
- Fiber build-out done on spec
- Predominantly non-investment-grade credits
- No contracted end-customer demand
- Turbine manufacturers had ~60 GW/yr capacity
AI Era
- Zero/near-zero spec builds
- AAA/AA+ credits (MSFT, GOOGL, AMZN)
- Hyperscalers have customers before ground breaks
- Manufacturers cautious - targeting 24-30 GW/yr
Equipment Winners (Per SemiAnalysis)
Scaling Fast
- Doosan - H-class entrant, 5 turbines for xAI
- Bloom Energy - 2GW/yr capacity by 2026
- Boom Supersonic - 2GW/yr by 2028
- Caterpillar - 2.5x turbine production by 2030
Playing Safe
- GE Vernova - Back to 24GW/yr (2007 levels)
- Siemens Energy - Price increases over capacity
- Wärtsilä - Protecting marine relationships
Next Bottlenecks
- Switchgear: Powell, Eaton, Schneider
- Midstream Gas: ET, WMB, TRGP
- EPCs: Argan (AGX) well-positioned