The AI trade could shift back in Nvidia's favor. Here's why - CNBC

As artificial intelligence continues to advance rapidly, global demand for high-performance computing chips is rising steadily. NVIDIA, as a leading player in the AI chip market, is now facing renewed... Read original →

NVIDIA’s strategic leap to 3nm EUV is triggering a full-stack cascade: upstream EDA and IP vendors are fast-tracking architecture-specific optimizations, while cloud providers scramble to retool software stacks for Blackwell’s power efficiency. Although TSMC’s expanded 3nm capacity in Taiwan, China eases near-term supply constraints, U.S. CHIPS Act 'guardrails' could restrict future advanced-node investments there, inflating long-term manufacturing costs. As AMD scales MI300X and Google deploys TPU v5, NVIDIA’s CUDA moat forces rivals into a trap—chasing hardware parity while lagging in ecosystem maturity. Over the next 18 months, surging AI inference demand—especially in autonomous driving and edge AI—will amplify NVIDIA’s integrated hardware-software advantage, making customized SoCs bundled with proprietary toolchains the new barrier to entry and cementing its oligopoly.