▶Both sources attribute the prediction that NVIDIA will reach a $10 trillion valuation within five years to Jonathan Ross.Feb–Mar 2026
▶Ross consistently identifies supply chain components, specifically High Bandwidth Memory (HBM) and interposers, as the primary bottleneck for NVIDIA's GPU production, rather than its ability to manufacture the silicon dies themselves.Feb 2026
▶He repeatedly connects national AI ambitions to massive investments in energy infrastructure, citing China's plan for 150 new nuclear reactors and Japan's efforts to bring reactors back online as crucial for powering AI.Feb 2026
▶Ross's analysis suggests that the demand for AI inference is so high that it is currently compute-limited, stating that doubling the inference capacity for companies like OpenAI and Anthropic would nearly double their revenues in a month.Feb 2026
▶Ross asserts that NVIDIA's CUDA software moat is 'incorrect' for the inference market, a position that directly challenges the widely held industry view that CUDA creates significant, long-term customer lock-in across all AI workloads.
▶He claims hyperscalers build their own AI chips primarily as a negotiation tactic to gain leverage over NVIDIA on pricing and supply, which contrasts with the common narrative that these custom chips are intended to fully replace NVIDIA hardware at scale.Feb 2026
▶Ross posits that HBM suppliers are hesitant to increase supply because the current scarcity allows them to maintain very high profit margins, suggesting a deliberate market strategy rather than a pure technological or manufacturing limitation.Feb 2026
▶His statement that Groq can deliver chips in six months versus a more than two-year lead time for NVIDIA GPUs presents a dramatic contrast in supply chain agility that implicitly questions the operational efficiency of the market leader.Feb 2026
Not enough data for timeline
Sign up free to see the full intelligence report
Get started free