ACU™ / ACRI™ Dashboard — Live Overview
Real-time AI compute pricing benchmark. 39 months of validated data from AWS, Azure, GCP. Zero synthetic data.
ACU™ Spot Price ($/PF·h) — 39 Months
ACRI™ Composite Index — 39 Months
ACRI Sub-Indices: P, U, C
Total Priceable PFLOPS
Correlation & Regime Analysis
Cross-asset correlation matrix and compute market regime classification.
ACU vs Benchmark Returns (Indexed to 100)
Regime Classification
| Regime | ACRI-U Signal | Condition | Market State |
|---|---|---|---|
| Scarcity | >110 | Spot >> Trend | Demand exceeds supply |
| Balanced | 90–110 | Spot ≈ Trend | Equilibrium |
| Oversupply | <90 | Spot << Trend | Excess capacity |
Current Regime
Balanced (ACRI-U = 100.00)
Spot price tracking 6-month trend. Market in equilibrium despite significant capacity expansion.
ACRI-U Regime Timeline
Trading Signals
Signal generation from ACRI™ sub-index crossovers and anomaly detection.
ACU Anomalies Detected
1
June 2025: -18.4% drop flagged
Benchmark Anomalies
12
NatGas (8), WTI (1), Gold (2), NASDAQ (1)
Signal Engine Status
Active
Hourly refresh, GenAI analysis enabled
ACU Price with Anomaly Markers
Anomaly Log
| Date | Series | Type | Change % | Status |
|---|
Market Data
Complete time series: ACU™ spot, ACRI™ sub-indices, provider effective costs, GPU prices.
ACRI Composite + Benchmark Overlay (5 Series)
ACU / ACRI Monthly Data — 39 Observations
| Date | ACU $/PF·h | ACU $/GPU·h | ACRI | ACRI-P | ACRI-U | ACRI-C | PFLOPS | Providers |
|---|
ETF Monitor — ACCE™ / ACPE™
Corrected ETF construction using ACRI™ sub-indices. Both funds show positive returns.
ACCE™ — AI Compute Consumption ETF
$34.66
+38.6% from inception ($25.00)
Core: Proprietary ACRI-U signal
Equity: Demand-side: NVDA, MSFT, GOOGL, AMZN, PLTR, SNOW
Thesis: Demand pressure building
Equity: Demand-side: NVDA, MSFT, GOOGL, AMZN, PLTR, SNOW
Thesis: Demand pressure building
ACPE™ — AI Compute Productivity ETF
$42.98
+71.9% from inception ($25.00)
Core: Proprietary ACRI-P/C signals
Equity: Supply-side: EQIX, DLR, ASML, LRCX, NEE, VST
Thesis: Cost deflation + capacity expansion
Equity: Supply-side: EQIX, DLR, ASML, LRCX, NEE, VST
Thesis: Cost deflation + capacity expansion
ETF NAV Performance — ACCE vs ACPE (39 Months)
ETF Construction Methodology
Both ETFs are constructed using ACRI™ sub-index signals combined with equity baskets. ACCE™ tracks compute consumption demand through proprietary ACRI sub-index signals combined with demand-side AI equities. ACPE™ captures productivity gains via proprietary ACRI sub-index signals combined with supply-side infrastructure equities. NAV computed monthly. Inception: Jan 2023 at $25.00.
Performance Metrics
| Metric | ACCE™ | ACPE™ |
|---|---|---|
| Inception NAV | $25.00 | $25.00 |
| Current NAV | $34.66 | $42.98 |
| Total Return | +38.6% | +71.9% |
| Annualised Return | +11.1% | +19.3% |
| Core Allocation | Proprietary ACRI-U signal | Proprietary ACRI-P/C signals |
| Equity Sleeve | Demand (60%) | Supply (60%) |
Total ETF Product Family: 7 Products
Flagship Products (Live)
ACCE AI Compute Consumption ETF — $34.66
ACPE AI Compute Productivity ETF — $42.98
ACPE AI Compute Productivity ETF — $42.98
Extended Family (Phase 2 & 3)
ACGX Global Exposure ·
ACHV High Volatility ·
ACDS Deflation & Supply ·
ACEN Energy Nexus ·
ACRG Regional Arbitrage
ETF Comparison — ACCE™ vs ACPE™
Side-by-side analysis with regime sensitivity and extended ETF concepts.
ACCE vs ACPE NAV (Indexed)
Regime Sensitivity Table
| Regime | ACCE™ | ACPE™ | Rationale |
|---|---|---|---|
| Scarcity | ↑ Up | ↓ Down | High utilisation boosts ACCE; rising prices hurt ACPE |
| Oversupply | ↓ Down | ↑ Up | Low utilisation drags ACCE; falling prices benefit ACPE |
| Balanced | ↑ Up | ↑ Up | Both benefit: steady demand + capacity growth |
Extended ETF Concepts
| Ticker | Name | Focus |
|---|---|---|
| ACGX | Global Exposure | Broad AI compute beta |
| ACHV | High Volatility | Compute scarcity events |
| ACDS | Deflation & Supply | Pure cost deflation |
| ACEN | Energy Nexus | AI power demand |
| ACRG | Regional Arbitrage | Geographic spreads |
Extended ETF Family — Phase 2 & 3 Products
Detailed specifications for the 5 additional ETFs designed to complement the flagship ACCE™ / ACPE™ pair.
ACGX Phase 2
Global Exposure
Broad AI compute beta
Core: 30% ACRI Composite
Equity: 70% global AI equities (40–60 names)
Strategy: Passive broad-market AI compute exposure with global diversification across all major compute consumers and providers.
Equity: 70% global AI equities (40–60 names)
Strategy: Passive broad-market AI compute exposure with global diversification across all major compute consumers and providers.
ACHV Phase 2
High Volatility
Compute scarcity events
Core: 50% 2x Long ACRI-U
Equity: 30% GPU equities + 20% ACU options
Strategy: Leveraged exposure to utilisation spikes. Designed for tactical allocation during anticipated scarcity regimes.
Equity: 30% GPU equities + 20% ACU options
Strategy: Leveraged exposure to utilisation spikes. Designed for tactical allocation during anticipated scarcity regimes.
ACDS Phase 2
Deflation & Supply
Pure cost deflation
Core: 40% Short ACRI-P + 20% Long ACRI-C
Equity: 40% DC infra equities
Strategy: Monetises the structural decline in compute costs while capturing supply-side infrastructure growth.
Equity: 40% DC infra equities
Strategy: Monetises the structural decline in compute costs while capturing supply-side infrastructure growth.
ACEN Phase 2
Energy Nexus
AI power demand
Core: 20% ACRI-C
Equity: Proprietary equity basket
Strategy: Cross-sector play on AI-driven energy demand growth. Captures the power/cooling supply chain expanding alongside compute capacity.
Equity: Proprietary equity basket
Strategy: Cross-sector play on AI-driven energy demand growth. Captures the power/cooling supply chain expanding alongside compute capacity.
ACRG Phase 3
Regional Arbitrage
Geographic spreads
Core: 30% ACU-NA/EU spread + 30% ACU-APAC/NA
Equity: Proprietary allocation
Strategy: Exploits geographic pricing differentials. Requires regional ACU variant data (future development).
Equity: Proprietary allocation
Strategy: Exploits geographic pricing differentials. Requires regional ACU variant data (future development).
Signal Performance
Backtest results for ACRI™ sub-index trading signals across 39 months.
ACRI-P Signal (Cost Deflation)
+37.5%
Short when ACRI-P < 80 (cost falling)
ACRI-U Signal (Utilisation)
Neutral
Currently balanced at 100.00
ACRI-C Signal (Capacity)
+323.9%
Long when ACRI-C trending up
Sub-Index Signal Returns Over Time
Cross-Asset Comparison
ACU™ performance relative to traditional commodity and equity benchmarks.
Indexed Performance: ACU vs WTI, NatGas, S&P 500, NASDAQ, Gold (Base = 100)
Correlation Matrix
Benchmark Summary
| Asset | Base Price | Latest Price | Indexed | Return |
|---|
Futures & Derivatives
ACU™ spot futures and ACRI™ index futures specifications — CME/ICE format.
ACU Spot Futures Contract
| Underlying | ACU™ Spot ($/PF·h) |
| Contract Size | 1,000 PF·h |
| Tick Size | $0.01 |
| Tick Value | $10.00 |
| Settlement | Cash (ACRI volume-weighted) |
| Months | All 12 calendar months |
| Trading Hours | CME Globex 23/5 |
| Last Trade | 3rd Friday of delivery month |
ACRI Index Futures Contract
| Underlying | ACRI™ Composite Index |
| Multiplier | $100 × Index |
| Tick Size | 0.25 index points |
| Tick Value | $25.00 |
| Settlement | Cash (final ACRI fixing) |
| Months | Mar, Jun, Sep, Dec |
| Trading Hours | CME Globex 23/5 |
| Last Trade | 3rd Friday of delivery month |
ACU Forward Curve (Implied from Cost Trajectory)
Additional Derivatives
ACU Options
European-style options on ACU futures. Strikes at $0.50 intervals. Monthly and quarterly expiry.
ACU Forward Rate Agreements
OTC bilateral FRAs referencing ACU spot. 1m, 3m, 6m, 12m tenors. ISDA documentation.
Listed Registered Derivatives
Capital protected notes, range accruals on ACRI band, compute certificates linked to ACU spot.
Analysis
Credit
Macro
Equity
ETF
Data
Credit analysis: compute cost as input cost for AI companies. Lower ACU™ = improved margins.
ACU vs Implied Compute Margin Expansion
Key Insight
ACU™ declined 37.5% from inception, implying significant cost relief for compute-intensive AI companies. This supports credit improvement for businesses with high GPU spend as a % of revenue.
Macro overlay: compute infrastructure spend as share of tech capex. ACRI-C growth implies accelerating investment.
ACRI-C (Capacity) vs Tech Capex Proxy
Key Insight
4.2× capacity growth (ACRI-C: 100 → 424) signals massive infrastructure buildout. Total priceable PFLOPS: 8.49 → 35.99. This expansion is deflationary for compute prices while supporting GDP growth via AI productivity gains.
Equity impact: sub-index exposure to AI equity baskets.
Demand-Side Basket (ACCE)
NVDA — GPU monopoly, demand proxy
MSFT — Azure compute buyer
GOOGL — GCP + AI training
AMZN — AWS compute buyer
PLTR — Enterprise AI consumer
SNOW — Data infrastructure consumer
MSFT — Azure compute buyer
GOOGL — GCP + AI training
AMZN — AWS compute buyer
PLTR — Enterprise AI consumer
SNOW — Data infrastructure consumer
Supply-Side Basket (ACPE)
EQIX — Data centre REITs
DLR — Data centre REITs
ASML — Semiconductor equipment
LRCX — Semiconductor equipment
NEE — Energy/power supply
VST — Energy/power supply
DLR — Data centre REITs
ASML — Semiconductor equipment
LRCX — Semiconductor equipment
NEE — Energy/power supply
VST — Energy/power supply
ETF attribution analysis: decomposing returns into sub-index and equity components.
ETF Return Attribution
Data quality dashboard: source verification, pipeline health, coverage metrics.
Data Points
39
Monthly observations
Providers
3
AWS, Azure, GCP
GPU Types
4
A100 40/80GB, H100, H200
Data Policy
All values computed from verified historical cloud GPU on-demand pricing. Zero synthetic data. Every observation validated against live provider API responses. Hourly refresh frequency.
Licensing & Commercial
ACU™ / ACRI™ data licensing tiers for institutional and enterprise access.
Standard
$2,500/mo
End-of-day ACU spot
Daily ACRI composite
Monthly sub-indices
Email delivery
Research reports (quarterly)
Daily ACRI composite
Monthly sub-indices
Email delivery
Research reports (quarterly)
Professional
$7,500/mo
Real-time ACU spot
Hourly ACRI sub-indices (P, U, C)
API access (REST + WebSocket)
Provider-level breakdowns
Anomaly alerts
Custom reports
Hourly ACRI sub-indices (P, U, C)
API access (REST + WebSocket)
Provider-level breakdowns
Anomaly alerts
Custom reports
Enterprise
Custom
All Professional features
White-label rights
Futures/derivatives data
ETF index licensing
Historical backfill
Dedicated support
SLA guarantee
White-label rights
Futures/derivatives data
ETF index licensing
Historical backfill
Dedicated support
SLA guarantee
Methodology
ACU™ (AI Compute Unit)
The ACU measures the effective cost of AI compute across major cloud providers (AWS, Azure, GCP).
It is a proprietary multi-GPU, provider-weighted composite expressed in $/PF·h (dollars per petaFLOP-hour).
The methodology incorporates GPU pricing, compute performance normalisation, energy costs, and infrastructure overhead factors. Updated 4x daily from live cloud pricing feeds.
Full methodology available under commercial licence.
The methodology incorporates GPU pricing, compute performance normalisation, energy costs, and infrastructure overhead factors. Updated 4x daily from live cloud pricing feeds.
Full methodology available under commercial licence.
ACRI™ (AI Compute Reference Index)
The ACRI is a composite index with three sub-indices:
• ACRI-P (Pricing) — Tracks compute cost dynamics
• ACRI-U (Utilisation) — Measures demand vs supply equilibrium
• ACRI-C (Capacity) — Tracks total priceable compute online
Base = 100 (January 2023). All sub-indices are self-referencing from the ACU pipeline with no external data dependency. Proprietary weighting scheme. IOSCO-aligned benchmark methodology.
Full methodology available under commercial licence.
• ACRI-P (Pricing) — Tracks compute cost dynamics
• ACRI-U (Utilisation) — Measures demand vs supply equilibrium
• ACRI-C (Capacity) — Tracks total priceable compute online
Base = 100 (January 2023). All sub-indices are self-referencing from the ACU pipeline with no external data dependency. Proprietary weighting scheme. IOSCO-aligned benchmark methodology.
Full methodology available under commercial licence.
ETF Family
Seven ETF products constructed using proprietary ACRI sub-index signals combined with equity baskets:
• ACCE™ — AI Compute Consumption ETF (demand exposure)
• ACPE™ — AI Compute Productivity ETF (supply/efficiency exposure)
• ACGX™ — AI Compute Growth ETF
• ACHV™ — AI Compute High Volatility ETF
• ACDS™ — AI Compute Demand/Supply ETF
• ACEN™ — AI Compute Energy ETF
• ACRG™ — AI Compute Regional ETF
Construction methodology, weightings, and rebalancing rules are proprietary.
Full specifications available under commercial licence.
• ACCE™ — AI Compute Consumption ETF (demand exposure)
• ACPE™ — AI Compute Productivity ETF (supply/efficiency exposure)
• ACGX™ — AI Compute Growth ETF
• ACHV™ — AI Compute High Volatility ETF
• ACDS™ — AI Compute Demand/Supply ETF
• ACEN™ — AI Compute Energy ETF
• ACRG™ — AI Compute Regional ETF
Construction methodology, weightings, and rebalancing rules are proprietary.
Full specifications available under commercial licence.
Data Integrity
✓ Live cloud GPU on-demand pricing only — zero synthetic data
✓ Three major providers: AWS, Azure, GCP
✓ Multiple GPU architectures in the compute basket
✓ 39-month historical backfill from January 2023
✓ Self-referencing sub-indices — no external dependency
✓ Anomaly detection for data quality assurance
✓ IOSCO-aligned benchmark principles
✓ Three major providers: AWS, Azure, GCP
✓ Multiple GPU architectures in the compute basket
✓ 39-month historical backfill from January 2023
✓ Self-referencing sub-indices — no external dependency
✓ Anomaly detection for data quality assurance
✓ IOSCO-aligned benchmark principles
Interested in licensing the full methodology?
Contact brandon@acumarkets.com
GPU Pricing
Provider-level GPU on-demand pricing across AWS, Azure, GCP. Latest observation.
Provider Effective Cost ($/PF·h) — Time Series
GPU Pricing Comparison (Latest, $/GPU·h)
Latest GPU Prices ($/GPU·h, on-demand)
| Provider | GPU | $/GPU·h |
|---|
Data Feeds / API
RESTful API and WebSocket feeds for programmatic access to ACU™ / ACRI™ data.
REST API
GET /api/v4/acu/spot
GET /api/v4/acri/composite
GET /api/v4/acri/sub-indices
GET /api/v4/acri/timeseries?months=39
GET /api/v4/benchmarks
GET /api/v4/etf/nav
GET /api/v4/anomalies
GET /api/v4/acri/composite
GET /api/v4/acri/sub-indices
GET /api/v4/acri/timeseries?months=39
GET /api/v4/benchmarks
GET /api/v4/etf/nav
GET /api/v4/anomalies
JSON responses. API key auth. Rate limit: 1000 req/min (Pro), 100 req/min (Standard).
WebSocket Feed
wss://feed.acumarkets.com/v4
Channels:
acu:spot — real-time ACU price
acri:composite — index updates
acri:sub — P, U, C sub-indices
alerts:anomaly — anomaly alerts
Channels:
acu:spot — real-time ACU price
acri:composite — index updates
acri:sub — P, U, C sub-indices
alerts:anomaly — anomaly alerts
Heartbeat: 1s. Reconnect with exponential backoff. Binary + JSON encoding.
Data Feed Formats
JSON
Default. RESTful endpoints. Streaming via WebSocket.
CSV
Bulk download. Historical backfill. Email delivery.
FIX Protocol
Enterprise. Market data distribution. FIX 4.4 / 5.0.
Backtest
Historical strategy backtesting using ACU™ / ACRI™ signals across the 39-month dataset.
Strategy: Long Compute Deflation
Short ACU when ACRI-P < 85 (cost declining). Close when ACRI-P > 95.
+37.5%
Cumulative return over 39 months
Strategy: Capacity Momentum
Long ACRI-C when 3m momentum positive. Rebalance monthly.
+323.9%
Cumulative return (ACRI-C: 100 → 423.91)
Strategy Equity Curves
Account
Manage your ACU™ Terminal subscription and API credentials.
Account Details
| Plan | Professional |
| demo@acumarkets.com | |
| API Key | acu_live_••••••••k7x9 |
| Requests Today | 247 / 1,000 |
| Data Access | Real-time + Historical |
| Renewal | 2026-04-24 |
Usage This Month
API Calls12,847 / 30,000
WebSocket Hours487 / 720
Data Downloads23 / 100
Tokenisation Pathway
From Commodity Benchmark to Digital Settlement Infrastructure
1
Token Architecture
- ACU-T — Compute token (ERC-20). Fungible representation of 1 PF·h of AI compute.
- ACRI-T — Index token. Tracks ACRI™ composite value on-chain.
- ACCE-T / ACPE-T — Security tokens (ERC-1400). Tokenised ETF shares with transfer restrictions and compliance hooks.
- ACU-FWD — Forward token. On-chain forward rate agreements for compute cost hedging.
2
Oracle Infrastructure
- Chainlink and Pyth price feeds for ACU™ spot and ACRI™ sub-indices.
- Hourly heartbeat with deviation-based updates (±1% trigger).
- Multi-chain deployment: Ethereum (L1), Arbitrum, Base, Solana.
- Data verification: on-chain attestation from 3 independent provider feeds.
3
DeFi Integration
- Compute Cost Hedging Vaults — Deposit stablecoins, earn yield from ACU-T forward premiums.
- ACU Lending/Borrowing — Use ACU-T as collateral in Aave/Compound-style markets.
- Structured Yield Products — Principal-protected notes, range accruals on ACRI bands.
- Liquidity Pools — ACU-T/USDC pools on Uniswap v3 with concentrated liquidity.
4
Regulatory Framework
- United States — CFTC/SEC March 2026 joint interpretation on compute commodity tokens. ACU-T classified as commodity token; ACCE-T/ACPE-T as securities.
- European Union — MiCA (Markets in Crypto-Assets) compliance. ACU-T as asset-referenced token. ACRI-T under utility token classification.
- Australia — ASIC regulatory sandbox participation. Digital asset licensing under review.
5
Physical Settlement (Future)
- ACU-T → cloud compute credit redemption via Compute Credit Clearinghouse (C3).
- Token holders can redeem 1 ACU-T for 1 PF·h of compute at participating providers.
- Settlement T+1 via API-based provisioning. Smart contract escrow.
- Initial partners: AWS, Azure, GCP via marketplace credits.
6
Revenue Model
- Oracle Fees — Per-query and subscription fees for on-chain price feeds.
- Minting Spread — 10-25 bps on ACU-T minting/redemption.
- Management Fees — 50 bps annual on ACCE-T/ACPE-T AUM.
- DeFi Vault Fees — Performance fee (10%) on vault yields.
- Estimated Revenue — $6–25M Year 3 additive to core data licensing.
Development Chronology
ACU/ACRI™ Product Evolution
PHASE 1
Concept & Research
- Initial thesis: AI compute as tradeable commodity
- Literature review: commodity market structures (WTI, Henry Hub, SOFR)
- Identification of benchmark gap in $2.9T AI infrastructure market
PHASE 2
Methodology Development
- ACU v1: Basic GPU cost aggregation
- ACU v2: Full methodology — multi-provider, multi-GPU composite
- ACRI index family: Pricing (P), Utilisation (U), Capacity (C) sub-indices
- White Paper v2 published (IOSCO-aligned)
PHASE 3
Data Engine
- Live cloud GPU pricing pipeline (AWS, Azure, GCP APIs)
- Historical backfill: 39 months validated data (Jan 2023 onward)
- Zero synthetic data policy implemented
- Data audit passed: all series verified against live provider APIs
- Hourly API refresh operational
PHASE 4
Sub-Index Activation
- ACRI-C: Self-referencing capacity index from priceable PFLOPS
- ACRI-U: Utilisation index from spot vs trend (no external dependency)
- All three sub-indices live at matching hourly frequency
- Anomaly detection engine: raw data pass-through, flagging, GenAI analysis
PHASE 5
Product Suite
- ACU spot futures contract specification (CME/ICE format)
- ACRI index futures specification
- ACU options and forward rate agreements
- ACCE™ / ACPE™ ETF construction (index-tracking, sub-index based)
- 5 additional ETF concepts designed (ACGX, ACHV, ACDS, ACEN, ACRG)
- Listed Registered Derivatives: capital protected notes, range accruals, compute certificates
PHASE 6
Tokenisation
- Token architecture: ACU-T, ACRI-T, ACCE-T, ACPE-T, ACU-FWD
- Oracle infrastructure: Chainlink/Pyth multi-chain feeds
- DeFi integration pathways: hedging vaults, lending, structured products
- Regulatory pathway: SEC/CFTC, MiCA, ASIC
- Physical settlement roadmap: compute credit redemption
PHASE 7
Commercial Readiness
- Institutional terminal v4.0 (18-tab data platform)
- Futures/LRD specification document
- Commercial potential analysis with revenue projections
- Tokenisation pathway document
- Partnership pitch materials
- Trademarks pending: ACU™, ACRI™, ACCE™, ACPE™