At Nvidia's annual GTC 2026 conference in San Jose, CEO Jensen Huang announced an optimistic outlook for the company's future revenue, projecting that purchase orders for its Blackwell and upcoming Vera Rubin systems will reach $1 trillion by the end of 2027. This figure represents a significant increase from last year's estimate of a $500 billion revenue opportunity for these two chip technologies. Following a recent earnings report that indicated growth exceeding expectations, Huang credited booming demand from both startups and major enterprises for this surge. The announcement helped push Nvidia shares up by approximately 2% on Monday, reinforcing its status as the world's most valuable public company with a market capitalization of roughly $4.5 trillion. The primary driver of this demand is the rapid shift in artificial intelligence adoption. As the industry moves from simple chatbots to agentic applications that generate complex tasks, the volume of data tokens being processed has exploded. Huang noted that if customers could secure more hardware capacity, they would immediately generate more tokens and see corresponding revenue increases. This need for speed and efficiency has been fueled by Nvidia's dominance in graphics processing units designed specifically for AI workloads. The company reported 11 consecutive quarters of revenue growth exceeding 55%, with year-over-year revenue for the current quarter projected to rise by 77% to approximately $78 billion. Looking ahead, Nvidia plans to release the Vera Rubin system later this year. This massive architecture, composed of 1.3 million components, is designed to deliver ten times the performance per watt of its predecessor, the Grace Blackwell. Addressing energy consumption as a critical challenge in AI infrastructure, Vera Rubin aims to provide a more sustainable solution for high-performance computing. In a major strategic move, Huang also unveiled the Groq 3 Language Processing Unit, or LPU. This chip represents Nvidia's first product derived from its $20 billion acquisition of Groq in December, the company's largest deal to date. Groq, founded by the creators of Google's tensor processing unit, specializes in low-latency computing. The new Groq 3 LPU is optimized to accelerate GPU performance, with a dedicated rack system, the Groq 3 LPX, capable of holding 256 LPUs. Huang stated that pairing this system with Vera Rubin could increase tokens per watt performance by 35 times, effectively combining high throughput with low latency while expanding memory capacity. Additionally, the CEO showcased a prototype of Kyber, the next evolution in rack architecture. Unlike current horizontal designs, Kyber integrates 144 GPUs in vertically oriented compute trays to boost density and reduce latency. This technology will be central to the Vera Rubin Ultra system, expected to ship in 2027. Together, these innovations underscore Nvidia's commitment to scaling AI infrastructure while addressing the physical and energetic constraints of modern data centers.