The primary NVIDIA Blackwell-powered knowledge middle GPU constructed for each enterprise AI and visible computing — the NVIDIA RTX PRO 6000 Blackwell Server Version — is designed to speed up essentially the most demanding AI and graphics purposes for each {industry}.
In comparison with the previous-generation NVIDIA Ada Lovelace structure L40S GPU, the RTX PRO 6000 Blackwell Server Version GPU will ship a multifold enhance in efficiency throughout a big selection of enterprise workloads — as much as 5x greater massive language mannequin (LLM) inference throughput for agentic AI purposes, almost 7x quicker genomics sequencing, 3.3x speedups for text-to-video era, almost 2x quicker inference for recommender techniques and over 2x speedups for rendering.
It’s a part of the NVIDIA RTX PRO Blackwell collection of workstation and server GPUs introduced at present at NVIDIA GTC, the worldwide AI convention going down via Friday, March 21, in San Jose, California. The RTX PRO lineup contains desktop, laptop computer and knowledge middle GPUs that assist AI and inventive workloads throughout industries.
With the RTX PRO 6000 Blackwell Server Version, enterprises throughout numerous sectors — together with structure, automotive, cloud companies, monetary companies, sport improvement, healthcare, manufacturing, media and leisure and retail — can allow breakthrough efficiency for workloads similar to multimodal generative AI, knowledge analytics, engineering simulation, and visible computing.
Content material creation, semiconductor manufacturing and genomics evaluation corporations are already set to harness its capabilities to speed up compute-intensive, AI-enabled workflows.
Common GPU Delivers Highly effective Capabilities for AI and Graphics
The RTX PRO 6000 Blackwell Server Version packages highly effective RTX AI and graphics capabilities in a passively cooled kind issue designed to run 24/7 in knowledge middle environments. With 96GB of ultrafast GDDR7 reminiscence and assist for Multi-Occasion GPU, or MIG, every RTX PRO 6000 could be partitioned into as many as 4 totally remoted situations with 24GB every to run simultaneous AI and graphics workloads.
RTX PRO 6000 is the primary common GPU to allow safe AI with NVIDIA Confidential Computing, which protects AI fashions and delicate knowledge from unauthorized entry with sturdy, hardware-based safety — offering a bodily remoted trusted execution setting to safe all the workload whereas knowledge is in use.
To assist enterprise-scale deployments, the RTX PRO 6000 could be configured in high-density accelerated computing platforms for distributed inference workloads — or used to ship digital workstations with NVIDIA vGPU software program to energy AI improvement and graphics-intensive purposes.
The RTX PRO 6000 GPU delivers supercharged inferencing efficiency throughout a broad vary of AI fashions and accelerates real-time, photorealistic ray tracing of advanced digital environments. It contains the most recent Blackwell {hardware} and software program improvements like fifth-generation Tensor Cores, fourth-generation RT Cores, DLSS 4, a totally built-in media pipeline and second-generation Transformer Engine with assist for FP4 precision.
Enterprises can run the NVIDIA Omniverse and NVIDIA AI Enterprise platforms at scale on RTX PRO 6000 Blackwell Server Version GPUs to speed up the event and deployment of agentic and bodily AI purposes, similar to picture and video era, LLM inference, recommender techniques, laptop imaginative and prescient, digital twins and robotics simulation.
Accelerated AI Inference and Visible Computing for Any Business
Black Forest Labs, creator of the favored FLUX picture era AI, goals to develop and optimize state-of-the-art text-to-image fashions utilizing RTX PRO 6000 Server Version GPUs.
“With the highly effective multimodal inference capabilities of the RTX PRO 6000 Server Version, our clients will be capable to considerably scale back latency for picture era workflows,” mentioned Robin Rombach, CEO of Black Forest Labs. “We anticipate that, with the server version GPUs’ assist for FP4 precision, our Flux fashions will run quicker, enabling interactive, AI-accelerated content material creation.”
Cloud graphics firm OTOY will optimize its OctaneRender real-time rendering utility for NVIDIA RTX PRO 6000 Blackwell Server Version GPUs.
“The brand new NVIDIA RTX PRO 6000 Blackwell Server Version GPUs unlock brand-new workflows that have been beforehand out of attain for 3D content material creators,” mentioned Jules Urbach, CEO of OTOY and founding father of the Render Community. “With 96 GB of VRAM, the brand new server-edition GPUs can run advanced neural rendering fashions inside OctaneRender’s GPU path-tracer, enabling artists to faucet into unimaginable new options and instruments that mix the precision of conventional CGI augmented with frontier generative AI expertise.”
Semiconductor tools producer KLA plans to make use of the RTX PRO 6000 Blackwell Server Version to speed up inference workloads powering the wafer manufacturing course of — the creation of skinny discs of semiconductor supplies which might be core to built-in circuits.
KLA and NVIDIA have labored collectively since 2008 to advance KLA’s physics-based AI with optimized high-performance computing options. KLA’s industry-leading inspection and metrology techniques seize and course of photographs by working advanced AI algorithms at lightning-fast speeds to seek out essentially the most important semiconductor defects.
“Primarily based on early outcomes, we count on nice efficiency from the RTX PRO 6000 Blackwell Server Version,” mentioned Kris Bhaskar, senior fellow and vice chairman of AI initiatives at KLA. “The elevated reminiscence capability, FP4 diminished precision and new computational capabilities of NVIDIA Blackwell are going to be notably useful to KLA and its clients.”
Boosting Genomics and Drug Discovery Workloads
The RTX PRO 6000 Blackwell Server Version additionally demonstrates game-changing acceleration for genomic evaluation and drug discovery inference workloads, enabled by a brand new class of dynamic programming directions.
On a single RTX PRO 6000 Blackwell Server Version GPU, Fastq2bam and DeepVariant — parts of the NVIDIA Parabricks pipeline for germline evaluation — run as much as 1.5x quicker in contrast with utilizing an L40S GPU, and 1.75x quicker in contrast with utilizing an NVIDIA H100 GPU.
For Smith-Waterman, a core algorithm utilized in many sequence alignment and variant calling purposes, RTX PRO 6000 Blackwell Server Version GPUs speed up throughput as much as 6.8x in contrast with L40S GPUs.
And for OpenFold2, an AI mannequin that predicts protein buildings for drug discovery analysis, RTX PRO 6000 Blackwell Server Version GPUs enhance inference efficiency by as much as 4.8x in contrast with L40S GPUs.
Genomics firm Oxford Nanopore Applied sciences is collaborating with NVIDIA to carry the most recent AI and accelerated computing applied sciences to its sequencing techniques.
“The NVIDIA Blackwell structure will assist us drive the real-time sequencing evaluation of something, by anybody, anyplace,” mentioned Chris Seymour, vice chairman of superior platform improvement at Oxford Nanopore Applied sciences. “With the RTX PRO 6000 Blackwell Server Version, we have now seen as much as a 2x enchancment in basecalling velocity throughout our Dorado platform.”
Availability through International Community of Cloud Suppliers and System Companions
Platforms that includes the RTX PRO 6000 Blackwell Server Version shall be out there from a worldwide ecosystem of companions beginning in Might.
AWS, Google Cloud, Microsoft Azure, IBM Cloud, CoreWeave, Crusoe, Lambda, Nebius and Vultr shall be among the many first cloud service suppliers and GPU cloud suppliers to supply situations that includes the RTX PRO 6000 Blackwell Server Version.
Cisco, Dell Applied sciences, Hewlett Packard Enterprise, Lenovo and Supermicro are anticipated to ship a variety of servers that includes the RTX PRO 6000 Blackwell Server Version, as are Advantech, Aetina, Aivres, ASRockRack, ASUS, Compal, Foxconn, GIGABYTE, Inventec, MSI, Pegatron, Quanta Cloud Expertise (QCT), MiTAC Computing, NationGate, Wistron and Wiwynn.
To be taught extra in regards to the NVIDIA RTX PRO Blackwell collection and different developments in AI, watch the GTC keynote by NVIDIA founder and CEO Jensen Huang: