The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →CoreWeave made NVIDIA HGX B300 generally available on CoreWeave Cloud on March 16, 2026. The offering is an eight-GPU instance aimed at demanding AI training and inference; CoreWeave’s published Spot rates are $35.84 per hour in North America and $36.70 per hour in Europe, while on-demand pricing requires contacting sales. Those are live, changeable rates for the full instance—not a quoted per-GPU price.
What is NVIDIA HGX B300 on CoreWeave?
HGX B300 is a hosted GPU-compute configuration available through CoreWeave Cloud. CoreWeave’s March 16, 2026 announcement describes it as a platform for reasoning, inference, long-context work, and training models with 100B+ parameters on a node. These are vendor-stated target capabilities, not independent benchmark results. CoreWeave’s launch announcement also says the system has 2.1 TB of HBM3e memory—50% more than its HGX B200 instances—and supports next-generation Quantum-X800XDR InfiniBand, which the company says doubles node-to-node bandwidth.
| # | Preview | Product | Price | |
|---|---|---|---|---|
| 1 |
|
NVIDIA Tesla A100 Ampere 40 GB Graphics Processor Accelerator - PCIe 4.0 x16 - Dual Slot | $4,669.00 | Buy on Amazon |
| 2 |
|
Nvidia RTX A1000 | Buy on Amazon | |
| 3 |
|
NVIDIA NVS 510 Graphics Card 0B47077 | $69.00 | Buy on Amazon |
| 4 |
|
NVIDIA Quadro RTX8000 (Renewed) | Buy on Amazon | |
| 5 |
|
NVIDIA RTX A1000 8GB ATX | $600.00 | Buy on Amazon |
As an Amazon Associate I earn from qualifying purchases.
The figures use different scopes: the launch release gives 2.1 TB HBM3e for the HGX B300 system, while CoreWeave’s instance documentation lists 270 GB of GPU RAM per GPU. They should not be treated as interchangeable measurements. The memory increase and bandwidth statements are CoreWeave’s claims; the announcement does not establish an independent performance comparison with B200.
What does a CoreWeave B300 instance include?
CoreWeave documents the instance as b300-8x, or eight NVIDIA B300 GPUs. Its listed configuration is:
#1 Best Overall
- Discrete graphics card memory 40 GB
- Memory bandwidth (max) 1555 GB/s
- Graphics processor family NVIDIA
- Graphics processor A100
| Component | Documented specification |
|---|---|
| GPUs | 8 NVIDIA B300 |
| GPU RAM | 270 GB per GPU |
| CPUs | 2 Intel Granite Rapids 6747P CPUs at 2.70 GHz |
| vCPUs | 192 |
| System RAM | 4,096 GB |
| Usable storage | 30.72 TB, configured as eight 7.68 TB devices in RAID 10 |
| Networking | Dual-port 200GbE, Quantum-X800 InfiniBand, and NVLink |
| Default GPU driver | 595; compatible driver versions listed: 580 and 595 |
These specifications are from the CoreWeave B300 instance documentation, last modified June 23, 2026. The page identifies the instance as “B300 (InfiniBand).”
How much does a CoreWeave B300 instance cost?
CoreWeave’s pricing page lists the following Spot rates for its eight-GPU HGX B300 instance, as accessed September 30, 2026:
Rank #2
- 3rd generation Tensor core and 2nd generation RT core provide 1.5 times more graphic CAD performance and 3 times more rendering and generating AI performance than previous model T1000
- It delivers up to twice the real-time lay-tracing performance of previous generations, allowing you to perform complex 3D model processing and more realistic image processing in half the time
- Up to 3.6 times higher generation AI performance than previous generations, creating high-quality images/videos, and generating 3D assets quickly
- Equipped with 4 Mini DisplayPort connectors for increased productivity for multi-application workflow. 8K output can also output 2 screens simultaneously
| Region | HGX B300 Spot rate | On-demand price |
|---|---|---|
| North America | $35.84 per hour | Contact sales |
| Europe | $36.70 per hour | Contact sales |
The figures are listed on CoreWeave’s live pricing page and can change; they are not guaranteed future rates. The page’s separate single-GPU inference-price layout is irregular for this entry, so it does not support a reliable per-GPU rate. Budgeting should account for the full node rate and the duration and availability of Spot capacity, rather than dividing the published figure and assuming that is a purchasable single-GPU price.
Recommended Free Tools
Where is CoreWeave B300 available?
The instance documentation names four US locations: US-CENTRAL-06A, US-EAST-13A, US-WEST-01A, and US-WEST-10A. CoreWeave’s pricing page also lists Spot pricing for North America and Europe. The pricing regions and the documentation’s named availability locations are not the same type of listing; check CoreWeave for current deployment availability before planning a workload.
Rank #3
- Quad display support1
- Display Port 1.2 features.H.264 Encoder
- Versatile connectivity options using Mini Display Port (mDP) connector.NVIDIA FXAA and TXAA.Intelligent Power Management
- Multi-display experience with NVIDIA Mosaic technology.NVidia High-Definition Video Technology
- Maximum Power Consumption:35Watts
What workloads is B300 intended for?
CoreWeave’s documentation names training foundation models from scratch at 130B+ parameters, massive-scale inference, and frontier AI development as primary uses. Its launch announcement additionally emphasizes reasoning and long-context workloads and describes training 100B+ parameter models on a single node. These use cases reflect CoreWeave’s product positioning; actual suitability depends on the model, software stack, memory needs, throughput target, and the availability and cost of the instance.
What else did CoreWeave announce?
The B300 launch was paired with Weights & Biases workflow capabilities for reinforcement learning and agent development. CoreWeave described Serverless RL as environment-free training and announced production-agent evaluation capabilities. These software features are distinct from the B300 hardware offering.
Rank #4
CoreWeave reported Weights & Biases claims of 1.4x faster training and up to 40% lower cost for Serverless RL compared with self-managed clusters. It also reported up to 5x lower inference cost and up to 60x lower latency, conditioned on no loss of quality. These are company-reported claims in the launch announcement, not independently validated results or guarantees for a particular workload.
What to compare before choosing B300
For an infrastructure decision, compare more than the GPU name. Relevant factors include GPU count and whether memory figures are per GPU or system-wide; interconnect and node-to-node bandwidth; workload type; region; Spot versus on-demand access; and the full node’s CPU, system memory, and storage. CoreWeave’s release compares B300 memory with B200, but does not provide independent, apples-to-apples performance results. Confirm current regional availability and pricing directly with CoreWeave before estimating spend.
Quick Recap
Best Value
- 900-5G172-2280-000
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
