NVIDIA A100 · 80 GB HBM2e · Ampere

Rent an A100 80 GB server.
$355.99 a month, no KYC.

A dedicated NVIDIA A100 with 80 GB of HBM2e at ~2.0 TB/s, with 16 vCPU, 128 GB of RAM and 3.84 TB NVMe for datasets and checkpoints. The value pick for fine-tuning and for serving large models, at about $0.488 an hour around the clock. Paid in crypto.

80 GB HBM2e, ~2.0 TB/s
3.84 TB NVMe for data
Ready in 1 to 24 hours
No KYC, crypto only

See the server What fits in 80 GB

01Plans & pricing

One A100 80 GB server, 3 jurisdictions.

The GPU is yours alone, with NVMe storage and an unmetered 1 Gbps port. Quarterly billing saves 5%, yearly 12%.

MonthlyQuarterly−5%Yearly−12%

GPU Servers: 1 plans, monthly prices in USD
GPU	VRAM	CPU	RAM	Storage	Locations	Price	Order
A100 80 GB Training and fine-tuning 80 GB VRAM16 vCPU128 GB RAM3.84 TB NVMe	80 GB	16 vCPU	128 GB	3.84 TB NVMe		$355.99/mo Cheapest competitor ~~$509.00~~	Deploy A100 80 GB

RTX 4090, 5090, L40S, A100, H100
CUDA drivers preinstalled
Single and multi-GPU nodes
No KYC · BTC, ETH, XMR, USDT, SOL

Need more bandwidth or FP8? The H100 has the same 80 GB at 3.35 TB/s.

02Specifications

The A100 80 GB, in numbers.

NVIDIA’s specifications for the card, and what comes with it in our server.

The A100 80 GB, in numbers.
Specification	A100 80 GB server
Architecture	NVIDIA Ampere
CUDA cores	6,912
GPU memory	80 GB HBM2e
Memory bandwidth	~2.0 TB/s
Tensor cores	3rd generation: TF32, BF16, FP16, INT8
FP16 / BF16 tensor (dense)	312 TFLOPS
FP64	9.7 TFLOPS, 19.5 with tensor cores
CPU, RAM, storage	16 vCPU · 128 GB · 3.84 TB NVMe
Network	1 Gbps, unmetered
Price	$355.99/mo, about $0.488 per hour

03Sizing

What runs on 80 GB.

The highest precision at which a dense language model fits with an 8K-token context and 10% headroom, the figures of our VRAM guide.

What runs on 80 GB.
Model size	A100 (80 GB)	L40S (48 GB)
7–8B	16-bit	16-bit
13–14B	16-bit	16-bit
32B	16-bit	8-bit
70B	4-bit	Does not fit
123B	Does not fit	Does not fit

LoRA fine-tuning of models up to about 14B fits in 80 GB; a full fine-tune of a 7–8B model needs about 16 bytes per parameter, so two H100. How we size models

04Monthly or hourly

When a month beats paying by the hour.

At $355.99 a month, the A100 costs about $0.488 an hour if it runs non-stop. Against an hourly rental, the month pays for itself after this much use:

When a month beats paying by the hour.
If the hourly rate is	Our month equals	Break-even use
$1.50 per GPU-hour	237 h	33% of the month
$2.00 per GPU-hour	178 h	24% of the month
$3.00 per GPU-hour	119 h	16% of the month

Hourly rates are examples, not quotes from any provider. We bill monthly, quarterly (−5%) or yearly (−12%), from your balance.

05Use cases

What an A100 is best at.

80 GB on a single card, at the lowest monthly price for that much memory here.

Fine-tuning
LoRA and QLoRA runs on private data that never leaves the server.
GPU images
Large models on one card
A 70B model at 4-bit, or a 32B model at 16-bit, without splitting it across GPUs.
vLLM guide
Data-heavy jobs
Embeddings and batch inference over large datasets, with 3.84 TB NVMe next to the GPU.
A100 plan
Scientific computing
FP64 at 9.7 TFLOPS, 19.5 with tensor cores, for simulations that consumer cards run slowly.
All GPU servers

06Jurisdictions

Where A100 servers run.

The same price in each of these jurisdictions.

Compare jurisdictions

Latency figures are estimates from distance, not measurements. Every jurisdiction has the same prices. How we estimate latency

07Offshore, in writing

The rules, before you pay.

What we promise is written into our policies, not just our marketing.

US DMCA noticesNot actioned
Answered with our policy, never enforced. Only a local court order, or a valid EU notice in EU locations, can require action.
DMCA policy
IdentityEmail only
No name, address, phone or ID document, ever. A private or disposable address is fine.
Privacy policy
Payment5 cryptocurrencies
Bitcoin, Ethereum, Monero, USDT and Solana, paid on-chain from any wallet to your balance. No card processor, no chargebacks.
Crypto payments
TransparencySigned canary
A PGP-signed warrant canary every quarter and a public count of every request we receive.
Warrant canary

A100 or H100?

Both have 80 GB. The H100 is faster, with about 1.7 times the bandwidth and FP8; the A100 costs less per month and is enough for fine-tuning and steady inference.

Does the A100 support FP8?

No. FP8 arrived with the H100. The A100 runs TF32, BF16, FP16 and INT8, and quantized models such as 4-bit and 8-bit formats run well on it.

Can I rent it by the hour?

No. Billing is monthly, quarterly (−5%) or yearly (−12%), paid from your balance. At about $0.488 an hour around the clock, a month costs less than hourly rentals once you use the card for a few hours a day.

How fast is delivery?

Between 1 and 24 hours after you order, depending on the jurisdiction. The access details appear on the server’s page in your client area.

Are the driver and CUDA installed?

Yes, on the Ubuntu 24.04 image. The GPU images guide covers PyTorch, Ollama, vLLM and ComfyUI.

Can I get a refund?

GPU servers are not refundable once delivered, because the hardware is set aside for you. Check the specifications above before you order.

80 GB of GPU memory, offshore.

1Pick a jurisdiction
2Pay in crypto, no ID asked
3SSH in within 1 to 24 hours

Configure your A100 All GPU servers

$355.99/mo · dedicated GPU · CUDA preinstalled

Rent an A100 80 GB server.
$355.99 a month, no KYC.

One A100 80 GB server, 3 jurisdictions.

The A100 80 GB, in numbers.

What runs on 80 GB.

When a month beats paying by the hour.

What an A100 is best at.

Fine-tuning

Large models on one card

Data-heavy jobs

Scientific computing

Where A100 servers run.

The rules, before you pay.

A100 servers, answered.

80 GB of GPU memory, offshore.

Welcome back

Rent an A100 80 GB server. $355.99 a month, no KYC.

Fine-tuning

Large models on one card

Data-heavy jobs

Scientific computing

80 GB of GPU memory, offshore.

Rent an A100 80 GB server.
$355.99 a month, no KYC.