Nvidia CEO Jensen Huang announced at the company's annual GTC developer conference that they anticipate purchase orders for their Blackwell and Vera Rubin chip platforms to hit $1 trillion by 2027, double their previous forecast. The surge in demand for AI infrastructure is being driven by startups and large enterprises, with Huang emphasizing that compute constraints are currently limiting industry growth. This optimistic projection reflects the shift towards more advanced AI applications that can autonomously perform tasks, leading to a significant increase in token generation.
Nvidia's bullish outlook coincides with the evolution of AI usage from basic chatbots to more sophisticated applications that require faster and more efficient inference. The company's GPUs continue to dominate this space, with Nvidia expecting its revenue to soar by approximately 77 percent year-on-year to about $78 billion in the current quarter. Nvidia's upcoming Vera Rubin system, set to launch later this year, is poised to deliver ten times more performance per watt than its predecessor, addressing the growing concern around energy consumption in scaling AI infrastructure globally.
During the conference, Huang also introduced the Groq 3 Language Processing Unit (LPU), Nvidia's first major product following the acquisition of the startup in a $20 billion deal. With a core designed to accelerate processing speeds, the Groq 3 chip will complement Nvidia's GPUs. Additionally, Nvidia unveiled a new Groq 3 LPX rack system capable of housing 256 LPUs, enhancing the tokens-per-watt performance of Rubin GPUs by up to 35 times. Nvidia's focus on optimizing both performance and physical infrastructure underscores the company's commitment to meeting the demands of increasingly complex AI workloads.
Looking ahead, Nvidia presented a prototype of its next-generation rack architecture, Kyber, which will integrate 144 GPUs arranged vertically to enhance density and reduce latency. Kyber, part of the Vera Rubin Ultra platform expected to launch in 2027, underscores Nvidia's drive to optimize AI infrastructure as workloads become more resource-intensive. Moreover, Nvidia's commitment to the AI ecosystem was highlighted through the introduction of the open-source AI framework 'OpenClaw' and the developer stack NemoClaw, positioning Nvidia as a comprehensive AI platform provider. Beyond data centers, Nvidia is expanding its presence in automotive AI, with partnerships with companies like Uber and automakers such as Nissan, BYD, Geely, Isuzu, and Hyundai to develop autonomous vehicles and buses powered by Nvidia's technology.