Arsenale Logo

IntelligenceOS

CoverageOrnn Compute Exchange
GPU Compute Price TrackerThree-Sided Demand SignalsUS Data Center Buildout MapFederal Compute & Data Center PolicyAI Data Center Build CycleHardware Cycle

Ornn Compute Exchange

Fintech / Capital Markets

Founders

Kush Bavaria, Wayne Nelms

Stage

Seed / $5.7M

ornn.com

Financial infrastructure for GPU compute — price indices, futures, swaps, and residual value protection.

Co-Investors

Crucible Capital, Vine Ventures, Link Ventures, BoxGroup

Key Updates

11 days ago
  • Academic research now shows compute futures can reduce enterprise cost volatility by 62-78% while financial institutions are actively deploying GPU-collateralized debt structures with investment-grade ratings, providing the first concrete validation that all three customer segments have genuine demand beyond theoretical interest.
  • H100 pricing has collapsed from above $7/hr in early 2024 to around $2.95/hr currently while maintaining extreme volatility and fragmentation ($1.03-$10/hr range across providers), directly validating the core thesis that GPU pricing chaos creates demand for hedging instruments.
  • The Trump administration's coordinated federal push to accelerate data center buildout through FERC jurisdiction assertions and streamlined permitting represents a dramatically more favorable regulatory environment for both exchange licensing and the underlying compute market growth that Ornn depends on.
  • Multiple competitors are now operating in the compute derivatives space including DRW-backed Compute Exchange projecting a $5 trillion market and CFTC-regulated Pluto, indicating the market is maturing but also that Ornn faces real competition for market share.
  • NVIDIA's Blackwell B200 general availability with concrete pricing expectations of $2.50-3.00/hour by Q4 2026 creates clear hardware depreciation timelines that directly support Ornn's residual value swap products as upgrade cycles compress.

Generated 11 days ago

GPU Compute Price Tracker

11 days ago
HIGH VOLATILITY
GPU$/hr RangeTrend
H100$1.03 – $10.00
declining
H200$3.00 – $3.59
increasing
B200$2.12+—
A100$0.35 – $2.70
stable
MI300X——
Price-Moving Events
AWS increased H200 instance pricing by 15%
H100 median pricing collapsed from $7+/hr to $2.95/hr since early 2024
Spot pricing offers 40-70% discounts but adds uncertainty

GPU compute pricing continues to show extreme volatility and fragmentation across providers, with H100 rates ranging from $1.03/hr to over $10/hr depending on the platform. AWS quietly increased H200 instance prices by 15% while H100 median pricing has collapsed from above $7/hr in early 2024 to around $2.95/hr currently, demonstrating the kind of dramatic price swings that create demand for hedging instruments. The spread between spot and on-demand pricing offers 40-70% savings but creates additional uncertainty for enterprises trying to budget compute costs. This persistent pricing chaos across the fragmented GPU marketplace directly validates Ornn's value proposition of providing predictable pricing through derivatives.

US Data Center Buildout Map

11 days ago
Projects
Operational
Under Construction
Planned
|State Posture
Tailwind
Mixed
Headwind
Click a state for details

Regional compute capacity divergence is accelerating as state regulatory environments create increasingly fragmented development patterns. Nevada is experiencing explosive 950% capacity growth with 3,800 MW in pipeline projects, while states like Maine implement construction moratoria despite gubernatorial vetoes, creating stark regional availability differences. Major hyperscale deployments like Meta's $10+ billion Lebanon campus and CleanArc's 900MW Virginia facility are concentrating capacity in permissive jurisdictions, while the overall pipeline shows 950 planned projects adding 332,151 MW of capacity across highly uneven geographic distribution. This regulatory fragmentation and concentrated buildout pattern is creating exactly the type of regional pricing disparities and capacity imbalances that Ornn's compute indices are designed to track and monetize.

Federal Compute & Data Center Policy

11 days ago
FERC Data Center Jurisdiction Assertion2026-10-01

Federal regulatory standardization of data center oversight creates clearer compliance pathway for Ornn's exchange licensing.

Federal Power Rules Bypass Bill2026-05-21

Proposed Senate bill allowing AI data centers to bypass federal power rules would increase compute supply for Ornn's exchange.

DATA Act of 20262026-03-09

Congressional legislation specifically targeting data center development creates federal policy framework supporting Ornn's infrastructure thesis.

Trump AI Executive Order2025-12-01

New AI policy framework that could standardize compute infrastructure requirements benefiting Ornn's exchange model.

Trump Executive Order on Federal Data Center Permitting2025-07-01

Accelerates federal permitting and revokes Biden-era climate requirements, removing regulatory barriers for Ornn's market expansion.

The Trump administration has launched a comprehensive federal push to accelerate data center buildout through multiple coordinated actions: signing a new AI executive order, directing FERC to assert federal jurisdiction over data centers, issuing executive orders to streamline federal permitting on federal lands, and having DOE roll out technical assistance programs. Congressional activity is building momentum with the DATA Act of 2026 and proposed legislation allowing data centers to bypass federal power rules through off-grid infrastructure. For Ornn Compute Exchange, this represents a dramatically more favorable regulatory environment where federal agencies are actively facilitating rather than constraining compute infrastructure development, which should create tailwinds for exchange licensing and the underlying compute market growth that Ornn depends on. The signal has strengthened significantly with coordinated executive action across DOE, FERC, and the White House moving from policy statements to concrete regulatory changes.

AI Data Center Build Cycle

11 days ago

Data Center Investment Scale

Goldman AI Capex Projection$7.6T — 2026-2031
Data Center Firms Capex$750B — 2026
Microsoft-Crusoe Deal700MW — Texas
Capacity Under Construction23GW — Current

The AI infrastructure build-out is accelerating into a multi-trillion dollar cycle, with Goldman Sachs projecting $7.6 trillion in capex through 2031 and BloombergNEF tracking $750 billion in current data center firm spending with 23GW of capacity under construction. The compute derivatives market is rapidly maturing with multiple competitors emerging - Compute Exchange (DRW-backed) projects a $5 trillion derivatives market while Pluto operates as a CFTC-regulated platform, validating the institutional demand for GPU pricing infrastructure. Concrete deals like Microsoft's 700MW lease at Crusoe's Texas facility and the broader trend toward GPU-backed debt and structured finance demonstrate that compute is becoming a tradeable asset class. This represents a strengthening signal for Ornn as both the underlying compute supply (driving pricing complexity) and financial sophistication (driving derivatives demand) are expanding simultaneously.

Three-Sided Demand Signals

11 days ago
AI Companies (Demand Side)Mixed

Academic research shows 62-78% cost volatility reduction through futures, with general industry discussion of hedging needs but no named entities publicly committing yet.

Data Center Operators (Supply Side)Mixed

McKinsey research confirms operators are hesitant to invest at maximum capacity due to limited demand visibility, indicating need for revenue certainty tools.

Financial Institutions (Capital Side)Strong

Equipment finance firms and private credit funds are actively developing GPU underwriting frameworks, with investment-grade GPU-collateralized debt already deployed.

Academic research is now demonstrating that compute futures can reduce enterprise cost volatility by 62-78%, while financial institutions including equipment finance firms and private credit funds are actively developing GPU-backed lending frameworks. McKinsey research confirms data center operators remain hesitant to invest at maximum capacity due to limited demand visibility, creating clear need for revenue stabilization tools. The emergence of GPU-collateralized debt structures with investment-grade ratings shows financial institutions are seriously engaging with GPU risk management, though credit analysts remain concerned about the risks. This represents the clearest evidence yet that all three customer segments have genuine demand for compute financial products, with concrete financial structures already being deployed.

Hardware Cycle

11 days ago
NvidiaH100 ASPs declining, secondary market growth
Blackwell B200Shipping
→
Vera Rubin (H2 2026)Announced

Primary index constituent. H100→B200 transition driving RVS demand.

AMDunknown
MI300XShipping
→
MI350 (mid-2025)Announced

Growing adoption diversifies OCPI.

Intelunknown
Gaudi 3unknown
→
unknown (unknown)unknown

Limited market presence.

Google
TPU Trillium (v6)unknown
→
unknown (unknown)unknown

Sets compute price floor.

AWS
Trainium2unknown
→
unknown (unknown)unknown

Cloud pricing feeds into OCPI.

Microsoft
Maia 100unknown
→
unknown (unknown)unknown

Same dynamic as Google/AWS.

NVIDIA's Blackwell B200 has reached general availability with production ramping through 2026, while H100 pricing shows clear depreciation signals at $25,000 purchase/$2.69 hourly with declining ASPs as secondary market inventory grows. AMD is accelerating its roadmap with MI350 moving to mid-2025 and MI400 series targeting 2026, creating additional depreciation pressure on current-generation chips. B200 cloud pricing is expected to stabilize at $2.50-3.00/hour by Q4 2026, indicating material compute cost deflation that will require OCPI index updates. The signal is strengthening as multiple vendors are compressing upgrade cycles and H100-to-B200 transition timing is now concrete, directly validating Ornn's residual value swap thesis.