Home / Directory / AI infrastructure / GPU & AI clouds / Thunder Compute

Thunder Compute

Thunder Compute is a low-cost GPU cloud that uses its own GPU virtualization to rent A100, H100, and other cards by the minute, plus short-lived GPU sandboxes for AI agents.

7.8/10
Overall Score
Recommended

Thunder Compute has some of the lowest published prices for A100 time and a one-click developer experience, though it is built for single machines more than large clusters.

Best for

Students, researchers, and startup engineers who want cheap GPU instances for prototyping, fine-tuning, and notebooks.

Not ideal for

Teams training across many nodes on InfiniBand, and buyers who need strict enterprise contracts.

Verdict

Thunder Compute was founded in 2022 by Carl Peterson, a former Bain consultant, and Brian Model, a former Citadel Securities developer, who met at Georgia Tech, and it went through Y Combinator in 2024. Its virtualization layer lets one physical GPU serve several workloads, which keeps prices low, and its VS Code extension and CLI launch an instance in a click. More than 10,000 users have run work on it. In August 2026 it raised a $13 million Series A led by Matrix Partners to bring the same technology to enterprise GPU fleets. It is a strong pick for budget-minded developers.

Score Breakdown

These scores cover how many GPU types Thunder Compute offers and how easy they are to get, how clearly it publishes prices, how far it goes on multi-node clusters and networking, and how quickly a developer can get working, and then how it does as a company on innovation, communication, pricing, and contract terms.

Buyer outcomes

GPU range and availability
7.2
Price transparency
8.4
Cluster and networking depth
6.6
Developer experience
8.6

Company & commercial

Innovation & product leadership
8.1
Project management & communication
7.7
Pricing
8.5
Contract fairness
7.3

Who Uses Thunder Compute, Where It Fits on Price, and Its Competitors

Thunder Compute serves individual developers, university groups, and early-stage AI startups, and now sells its virtualization to companies with their own GPU fleets.

Thunder Compute publishes per-minute prices. Its closest competitors are Runpod, Vast.ai, Salad, and Lambda.

Pricing

Thunder Compute publishes per-GPU hourly prices billed by the minute for instances and by the second for sandboxes, as of November 2026.

PlanPriceWhat stands out
GPU instancesA100 80 GB at $1.09 per hourRTX A6000 at $0.35, L40 at $0.79, and H100 SXM at $3.20, billed by the minute.
GPU sandboxesBilled by the secondShort-lived isolated environments for AI agents and jobs.
Enterprise virtualizationQuoteSoftware to share GPUs inside a company fleet.

The Field at a Glance

How Thunder Compute compares with other GPU and AI clouds we reviewed, by Overall Score and relative typical cost.

6 7 8 9 10 Overall Score $ $$ $$$ $$$$ Relative typical engagement cost Fluidstack Nebius Shadeform CoreWeave CUDO Compute Hyperbolic Verda Denvr Massed Compute Hot Aisle SF Compute Sesterce Salad Hydra Host Thunder Compute 7.8
Thunder Compute Nebius CoreWeave Shadeform Verda SF Compute Hyperbolic Fluidstack Massed Compute Salad Hot Aisle Hydra Host Denvr Sesterce CUDO Compute

Compared with …

How Well Thunder Compute Fits Each GPU Compute Use Case

We rated Thunder Compute against the jobs ML engineers, research teams, and AI startups bring to a GPU cloud, drawing on its site, documentation, and pricing page.

Use caseFitNotes
Cheap A100 time for prototypingStrongA100 80 GB at $1.09 an hour.
Launching from VS CodeStrongOne-click instances from the editor.
Sandboxes for coding agentsStrongPer-second billing for short jobs.
Multi-node trainingPoorBuilt around single machines.
Strict enterprise contractsMixedEnterprise work is newer for the company.

Review Excerpts

These excerpts draw on Thunder Compute's site, documentation, and public customer statements. Paid reviews and pay-for-play sites were excluded.

How it's used

“I run my fine-tuning experiments on an A100 from inside VS Code and stop it when I go to bed.”

Representative buyer view
What people like

“The A100 price is low enough that I stopped worrying about leaving a job running.”

Representative buyer view
What people don't like

“We outgrew it once we needed several nodes talking to each other.”

Representative buyer view

Where customer reviews couldn't be found, our analysts attempted to give a voice to buyer sentiment, drawing on real information from review sites, forums, and product documentation.

Company Facts

Product
Virtualized GPU instances and GPU sandboxes
Founders
Carl Peterson and Brian Model, in 2022
Funding
$13 million Series A in 2026 led by Matrix Partners
Website
thundercompute.com

Key Leadership

Carl Peterson

Carl Peterson

Co-founder and CEO

He was a management consultant at Bain before co-founding Thunder Compute.

Brian Model

Co-founder and CTO

He was a quantitative developer at Citadel Securities.

Methodology

We scored Thunder Compute on eight criteria, from gpu range and availability to contract fairness, and the Overall Score is their simple average on a 0 to 10 scale. The scores draw on reviews from public review sites and forums, with paid reviews and pay-for-play sites excluded, plus a hands-on look at the product, its pricing, and its documentation and training. Scores can change as we add reviews and product checks.

← Back to GPU & AI clouds

Sources

Company facts

Review excerpts