Nvidia Path to $20 Trillion Valuation Hinges on AI Inference Shift

Nvidia (NVDA) may be positioned for an unprecedented ascent to a $20 trillion valuation by 2030, according to a new analysis from Beth Kindig of the I/O Fund. The projection suggests the chipmaker is currently undervalued relative to its historical price-to-sales (P/S) premiums and comparable AI hardware peers such as AMD and Broadcom. The bullish outlook is predicated on a fundamental shift in the AI lifecycle. While initial growth was driven by the training of large language models (LLMs), the next phase of expansion will be fueled by 'inference'—the process where AI models apply existing knowledge to generate real-time intelligence. As agentic systems proliferate, demand is expected to shift toward high-throughput intelligence and system-wide efficiency. Kindig’s financial framework applies a P/S multiple of 22 to a projected annual data-center revenue target of $930 billion. This aligns with CEO Jensen Huang's guidance of $1 trillion in cumulative sales from the Blackwell and Rubin chip architectures through 2027. Furthermore, analyst consensus for fiscal 2031 has climbed to $758 billion, roughly double the expectations from one year ago. Despite the rise of custom silicon designs, Nvidia's integrated ecosystem—combining GPUs with the CUDA software layer and dominant networking capabilities—is expected to maintain its position within enterprise infrastructure budgets. The transition to the inference era is viewed as a multiplier for the total addressable market, driving higher utilization and recurring software revenue.