NVIDIA DGX B300

NVIDIA DGX B300

NVIDIA DGX B300 is a dedicated AI solution designed to meet the computational demands of generative AI. Featuring NVIDIA Blackwell Ultra GPUs, the DGX B300 delivers up to 11× higher performance for inference and 4× higher performance for training compared to the previous generation H200. It is compatible with both NVIDIA MGX kits and standard enterprise racks, and with the complete NVIDIA AI Enterprise software suite, it simplifies and accelerates AI deployment, enabling any enterprise to develop products as a hyperscaler.

NVIDIA DGX B300 is expected ~Q4 2025.

8 GPU

Blackwell Ultra

2.3 TB

GPU memory

144

PFLOPS FP4

14 kW

Consumption

30 TB

Data storage

72

PFLOPS FP8

Hardware Software NVIDIA GPU Cloud Support

System specifications

The NVIDIA DGX™ B300 is a dedicated AI infrastructure solution designed for businesses that want to harness the full potential of generative AI. The rapid growth of AI adoption across industries is presenting companies with new challenges – complex integration, lack of experts, and managing power consumption and costs.

The DGX B300, with large graphics memory as the core building block of the NVIDIA DGX SuperPOD, offers 144 petaFLOPS of performance for inference and 72 petaFLOPS for training. With NVIDIA Blackwell Ultra GPUs and a complete software stack, it enables businesses to efficiently scale AI solutions without the need for complex infrastructure.

The new DGX B300 format can be deployed on its own, as a “DGX Pod” or as an HGX in OEM partners’ chassis.

Parametr	DGX B300
GPU	8 × NVIDIA Blackwell Ultra
CPU	2x Intel Xeon CPU
GPU paměť	2.3 TB
Výkon	72 PFLOPS FP8 trénování 144 PFLOPS FP4 inference
Networking	8x OSFP ports serving 8x single-port NVIDIA ConnectX-8 VPI Up to 800Gb/s NVIDIA InfiniBand/Ethernet 2x dual-port QSFP112 NVIDIA BlueField-3 Up to 800Gb/s NVIDIA InfiniBand/EthernetDPU
Management networking	1GbE onboard NIC with RJ45 1GbE RJ45 Host baseboard management controller (BMC)
Úložiště	OS: 2x 1.9TB NVMe M.2 Interní úložiště: 8x 3.84TB NVMe E1.S
Spotřeba	~14kW
Software	NVIDIA DGX OS / NVIDIA Mission Control / NVIDIA AI Enterprise / NVIDIA Base Command Manager Podporuje Red Hat Enterprise Linux / Rocky / Ubuntu
Podpora	Tříletá standardní hardware a software podpora

Blackwell Ultra Architecture

AI development is no longer just about bigger models and more data in pretraining, but is moving towards refinement and the ability to “think”. Scaling post-training improves models for specific tasks by“fine-tuning” them on specialized and synthetic data, increasing their accuracy and understanding of context. In addition, synthetic data has no volume limit, which not only allows for much higher quality models, but also significantly increases the demands on computing power.

A new approach is test-time scaling (long thinking), where the model actively thinks during inference, considers multiple options and continuously adjusts the answers. This brings AI closer to agent intelligence – systems capable of independent reasoning and solving more complex tasks. These new trends – post-training and test-time scaling – dramatically increase the demands on computing resources, processing speed, and interconnectivity between systems. Post-training can easily require up to 30× more computing power than traditional pre-training and test-time scaling up to 100× more than commonly used inference.

More about architecture

NVIDIA AI Inference performance development chart

That’s why NVIDIA developed Blackwell Ultra, a computing platform designed for the era of “AI reasoning” that covers training, post-training and test-time scaling. Blackwell Ultra enables massive AI model inference with greater intelligence, speed and efficiency at optimal operational and infrastructure costs. It will be a key component of the NVIDIA GB300 NVL72 system, a liquid-cooled rack solution that combines 36 Grace CPUs and 72 Blackwell Ultra GPUs into a single unit with a total NVLink throughput of 130 TB/s, acting as one giant GPU for extremely demanding AI jobs.

The NVIDIA DGX B300 is the first finished product from this new generation of AI systems and brings many benefits and effective solutions to current challenges and obstacles on the road to AI development.

More about DGX B300

Datasheet DGX B300

Software

Included with virtually every NVIDIA device, whether it’s a compact AI workstation or a supercomputer at the scale of an entire data center, is the extensive but intuitive NVIDIA AI Enterprise software suite. With the various features and settings that this software brings, you’ll have a perfectly optimized environment on your device for machine learning, data analytics, or working with AI models as they are trained, inferenced, and ultimately tested and deployed. Libraries and tools such as Cafe 2, Theano, TensorFlow, PyTorch or MXNet are also ready for these purposes, and more are no problem to get from the portal NVIDIA GPU Cloud (NGC). Thanks to careful tuning and optimization, such an environment provides up to 30% higher performance for Machine Learning applications compared to those deployed on NVIDIA hardware alone, according to NVIDIA.

More about NVAIE

NVIDIA NIM

NVIDIA NIM™ provides containers for self-hosting GPU-accelerated inference microservices for pre-trained and customized AI models across clouds, datacenters, RTX™ AI PCs and workstations. NIM microservices expose industry standard APIs for easy integration into AI applications, frameworks and workflows. They are built on optimized inference engines from NVIDIA and the community, including NVIDIA TensorRT™ and TensorRT-LLM. NIM microservices optimize response latency and throughput for every combination of foundation model and GPU.

More about NIM

NVIDIA NeMo

NVIDIA NeMo™ is an end-to-end platform for the development of custom generative AI, including language models (LLMs), computer vision language models (VLMs), retrieval models, video models, and AI for voice processing.

With NeMo, you can easily create a “data flywheel” to continuously optimize and improve AI agents with the latest information. NeMo accelerates the data flywheel by curating AI and human feedback, refining and evaluating models, and deploying with retrieval-augmented generation (RAG) to ensure agents consistently achieve peak performance.

More about NeMo

NVIDIA GPU Cloud (NGC)

NVIDIA GPU Cloud (NGC) represents the repository of the most used frameworks for machine learning and deep learning applications, HPC applications, or NVIDIA GPU cards accelerated visualization. Deploying these applications is a question minutes — copying a link of the appropriate Docker image from the NGC repository, moving it on the DGX system, and downloading and running the Docker container. The individual development environments – versions of all included libraries and frameworks, settings of environment parameters – are updated and optimized by NVIDIA for deployment on DGX systems.

NVIDIA NGC Catalogue

Key benefits of NVIDIA support

The strength of the NVIDIA solution is to support the entire system. Hardware support (in case of failure of any of the components) is a matter of course. Software support for the entire environment is critical if something does not work as intended. The customer has hundreds of developers ready to help. Support is part of NVIDIA DGX purchase. It is available for 3 – 5 years and can be further extended at the end of that period.

Repair by a technician at the customer’s site

You get professional service right at your doorstep. A technician will arrive on-site, perform a diagnosis, and make the repair quickly and efficiently.

Remote diagnostics

Quick remote assistance. Our specialists will connect to your device and perform diagnostics and repairs without the need for a personal visit.

Extended software support

Provides longer access to updates, bug fixes and software enhancements. Your device stays up-to-date and protected even after warranty.

Private NCC storage and software

Secure and private storage for your software and data. Quick access to download tools and documentation from within the NVIDIA environment.

Latest updates and upgrades

Get access to the latest software and firmware releases that improve performance, security and add new device features.

Direct contact to NVIDIA support

Don’t hesitate to communicate with the NVIDIA team of experts. Quick answers, technical assistance and advice tailored to your needs.

NVIDIA DGX price action

NVIDIA offers special programs for DGX systems and Tesla accelerators for EDU organizations or start-up companies. Thanks to the international collaboration between NVIDIA and IBM Global Financing, preferential financing in the form of operating leases is available for DGX models.

Get a quote

M Computers
represents NVIDIA
on the Czech market in the Enterprise
area of computing
accelerators and AI systems.

Last year, it was the first company in Central and Eastern Europe to receive the highest ELITE PARTNER status, as well as two NVIDIA AI Innovator and NVIDIA AI Champion awards.

NVIDIA Deep Learning Institute

The NVIDIA Deep Learning Institute (DLI) offers both online and hands-on training for developers, data scientists, and researchers looking to solve challenging problems with deep learning and accelerated computing.

More information

Testing

We have a wide range of NVIDIA GPU cards and other NVIDIA systems available to test performance and process demanding tasks.
If you are interested in our testing offer, please fill out this form.

Kamila Jeřábková

NVIDIA, AI solutions

+420 734 161 516

kamila.jerabkova@mcomputers.cz

NVIDIA DGX B300

8 GPU

2.3 TB

144

14 kW

30 TB

72

System specifications

Blackwell Ultra Architecture

Software

NVIDIA NIM

NVIDIA NeMo

NVIDIA GPU Cloud (NGC)

Key benefits of NVIDIA support

Repair by a technician at the customer’s site

Remote diagnostics

Extended software support

Private NCC storage and software

Latest updates and upgrades

Direct contact to NVIDIA support

NVIDIA DGX price action

M Computers
represents NVIDIA
on the Czech market in the Enterprise
area of computing
accelerators and AI systems.

NVIDIA Deep Learning Institute

Testing

Kamila Jeřábková

News

By bike, on foot, sustainably: M Computers in the Bike to Work Challenge

Farming the future with HPE GPU servers

April Running and Walking Challenge 2025: we help by moving

Our ecological footprint 2024: savings from recycling, energy from PV

About us

Our websites

For your IT world

NVIDIA DGX B300

8 GPU

2.3 TB

144

14 kW

30 TB

72

System specifications

Blackwell Ultra Architecture

Software

NVIDIA NIM

NVIDIA NeMo

NVIDIA GPU Cloud (NGC)

Key benefits of NVIDIA support

Repair by a technician at the customer’s site

Remote diagnostics

Extended software support

Private NCC storage and software

Latest updates and upgrades

Direct contact to NVIDIA support

NVIDIA DGX price action

M Computers represents NVIDIA on the Czech market in the Enterprise area of computing accelerators and AI systems.

NVIDIA Deep Learning Institute

Testing

Kamila Jeřábková

News

About us

Our websites

For your IT world

M Computers
represents NVIDIA
on the Czech market in the Enterprise
area of computing
accelerators and AI systems.