NVIDIA announces latest generation of AI chips and software at San Jose developer’s conference

  • NVIDIA reveals the latest AI chips and software at the San Jose conference.
  • Introduces Blackwell series of GPUs led by GB200 chip, boasting 20 petaflops.
  • GB200 facilitates larger and more complex AI models with transformer engine.
  • Collaboration with cloud giants like Amazon, Google, Microsoft, and Oracle to offer GB200 access.
  • NVIDIA Inference Microservice (NIM) extends the life of older GPUs for inference tasks.
  • NIM simplifies AI model deployment across various platforms.

Main AI News:

NVIDIA made waves at its San Jose developer’s conference with the unveiling of its newest AI chips and software solutions, positioning itself at the forefront of AI innovation. Central to this reveal is the introduction of the Blackwell series of AI graphics processors, headlined by the powerhouse GB200 chip, scheduled for release later this year. Complementing this hardware leap are innovative software solutions, including the groundbreaking NVIDIA Inference Microservice (NIM), aimed at optimizing the use of older NVIDIA GPUs for inference tasks.

The Blackwell Revolution: Redefining AI Processing Power

At the heart of NVIDIA’s latest offerings lies the Blackwell series of graphics processors, spearheaded by the GB200 chip. This cutting-edge technology represents a quantum leap in AI computing prowess, boasting a staggering 20 petaflops of AI performance. Contrasted with its predecessor, the Hopper H100, which delivered a mere 4 petaflops, the GB200 sets a new standard for processing power. Designed to empower AI companies to craft larger and more intricate models, the Blackwell GPU incorporates a transformer engine tailored for transformer-based AI technologies, such as those underpinning ChatGPT. NVIDIA’s commitment to advancing GPU architecture promises a substantial performance enhancement with each biennial iteration, ensuring continuous strides in AI capabilities.

The GB200’s architecture, which integrates two B200 Blackwell GPUs with an Arm-based central processor, underscores NVIDIA’s dedication to propelling AI model training and deployment to new heights. Noteworthy is the collaboration with cloud titans like Amazon, Google, Microsoft, and Oracle, who are poised to provide access to the GB200, heralding widespread adoption and fostering unparalleled AI advancements.

NIM: Closing the Gap in AI Deployment

In tandem with its hardware innovations, NVIDIA introduces NIM (NVIDIA Inference Microservice), a pivotal addition to its enterprise software suite. NIM addresses the challenge of harnessing older NVIDIA GPUs for inference tasks, maximizing the utility of existing GPU investments. By extending the lifespan of current GPU assets and simplifying AI model deployment across diverse platforms, from on-premises servers to cloud infrastructures and even GPU-equipped laptops, NIM streamlines the integration of AI solutions into existing workflows. NVIDIA’s holistic approach to innovation, encompassing both hardware and software enhancements, signifies a strategic evolution from chip provider to comprehensive platform enabler. Positioned to equip developers and enterprises with the requisite tools for pioneering AI applications, NVIDIA reinforces its status as a trailblazer in the AI landscape.


NVIDIA’s unveiling of its latest AI chips and software solutions signifies a significant leap forward in AI capabilities. The introduction of the Blackwell series, led by the powerful GB200 chip, not only boosts processing power but also paves the way for the development of larger and more intricate AI models. Collaborations with major cloud providers expand access to these technologies, fostering widespread adoption and driving AI advancements across industries. The launch of NVIDIA Inference Microservice (NIM) further enhances the accessibility and efficiency of AI deployment, underlining NVIDIA’s transition from a chip provider to a comprehensive platform enabler in the evolving AI landscape.