Financial Institutions
Banks, insurers, and investment funds use the indices to value AI infrastructure exposure, benchmark private compute assets, and underwrite GPU-backed contracts.
Silicon Data Indices — daily GPU price benchmarks
GPU Rental Price Indices — Daily Benchmarks Across the AI Compute Market
Silicon Data Indices give financial institutions, AI companies, and infrastructure providers a transparent, independent view of what GPU capacity actually costs — updated every day, across neo-clouds and hyperscalers.
- 95% of neo-cloud GPU providers
- 100% of major hyperscalers
- Daily via API & partners
The Silicon Index family
GPU rental price indices
H100 and A100 are published as two indices each — a Neo-Cloud index, covering neo-clouds and the marketplace, and a Hyperscaler index — so you can track the spread between them. Newer models publish a single standardized index while segment coverage builds.
H100 Rental Price Index
Bloomberg ticker: SDH100RT
Neo-Cloud$2.70▲ 1.1% (7D)
Hyperscaler$7.24▼ 0.4% (7D)
(As of Jul 15, 2026)
H200 Rental Price Index
Bloomberg ticker: --
Rental price$2.99▲ 1.7% (7D)
(As of Jul 15, 2026)
A100 Rental Price Index
Bloomberg ticker: SDA100RT
Neo-Cloud$1.64▼ 0.6% (7D)
Hyperscaler$3.71▲ 0.3% (7D)
(As of Jul 15, 2026)
B200 Rental Price Index
Bloomberg ticker: SDB200RT
Rental price$5.62▼ 5.7% (7D)
(As of Jul 15, 2026)
MI300X Rental Price Index
Bloomberg ticker: --
Rental price$2.50▲ 3.3% (7D)
(As of Jul 15, 2026)
More GPU indices coming soon.
Stay tuned!
Token & Memory indices
Price benchmarks beyond GPU rentals — LLM token and memory markets, expanding as coverage grows.
Why Silicon Data Indices
Built for institutional use
GPU pricing moves daily, but the market has no common reference point. Silicon Indices are the first benchmarks built for institutional use — transparent, independent, and actionable.
1
Market Transparency
The first daily benchmarks spanning the full GPU rental market, from neo-clouds to hyperscalers. Published openly, not gated behind a single vendor's pricing page.
2
Independent and Objective
Silicon Data does not sell GPUs, operate clouds, or broker capacity. Our indices are calculated by independent agents and validated against multiple data sources.
3
Built for Action
Structured for procurement decisions, contract benchmarking, financial reporting, and capacity planning. Every number is traceable to a defined methodology.
4
Distribution Across Platforms
Available via direct API and distributed through partners including Kaiko.
Who uses Silicon Indices
From trading desks to ML infrastructure teams
AI Companies & Infrastructure Buyers
AI labs, startups, and enterprise ML teams use the indices to negotiate supplier contracts, plan training budgets, and validate internal cost forecasts.
Infrastructure Providers
Neo-clouds, data center operators, and GPU lessors use the indices to price offerings competitively and communicate value to customers. Many pair index access with the Forward Curve to plan ahead.
Coverage across the GPU rental market
The Silicon Index coverage network spans the full rental supply chain — from large neo-cloud operators to every major hyperscaler.
95%
Neo-cloud GPU providers tracked
100%
Major hyperscalers covered
80%
Of the available global GPU rental market
Methodology
How We Build Our Indices
At Silicon Data, accuracy, transparency, and rigor are at the core of every index we create. Here's how we do it.
01
Deep Data Coverage
Our proprietary data network covers 80%+ of the available global GPU rental market today, and is expanding. We collect real-time GPU rental and private transaction data across:
- Cloud providers
- Colocation markets
- Brokered cluster sales
- Private rental platforms
02
Robust Standardization Methodology
We standardize these variables to create a like-for-like, comparable on-demand rate — ensuring every index reflects true market costs. Prices vary based on:
- Machine specs (GPU type, memory, CPU)
- Rental terms (hourly, daily, monthly)
- Platform performance (interconnect, cluster scale)
- Geolocation (data center region)
03
Daily Calculation and Independent Validation
- Daily Calculation. Prices are captured and processed every business day.
- Outlier Removal. Statistically robust methods filter out stale or non-representative data.
- Quality Control. Validation procedures ensure index integrity, aligning with industry best practices.
04
Distribution and Access
- Direct API access (e.g. ticker SDH100RT)
- Partnerships with Kaiko and other platforms
- Historical data archives available
Frequently Asked Questions (FAQ)
The Silicon GPU rental price index is a family of daily benchmarks tracking hourly rental prices for specific GPU models across the full rental market. Each index represents a standardized, like-for-like price drawn from neo-cloud providers, hyperscalers, colocation markets, and private rental platforms.
Every index is processed and published once per business day, reflecting the latest captured rental and transaction data.
Today the family covers H100, A100, B200, and MI300X rental price indices, plus the LLM Token Expenditure Index — with more models added as coverage expands.
Neo-cloud and marketplace capacity typically prices below hyperscaler on-demand rates. H100 and A100 are tracked across both segments, with each reading published separately so you can see the spread and how it moves.
Via the Silicon Data portal, direct API (e.g. ticker SDH100RT), and distribution partners including Kaiko.
We aggregate real-time rental and private transaction data across neo-cloud providers, hyperscalers, colocation markets, and private rental platforms, then standardize it into a like-for-like benchmark.
Get access to Silicon Data
Full access to every Silicon Index, the complete portal, and historical pricing — distributed via web and API.
Silicon Data Portal Access
7-day trial →Daily GPU and LLM market intelligence, built for institutional use.
- SiliconIndex™ Dashboard — all GPU, Memory & accelerator indices
- Market direction tracking
- PriceIQ predictive modeling
- Daily index values & 90-day historical pricing
- NeoCloud vs Hyperscaler comparison
- SiliconCarbon™