It’s pronounced en-vidia (Kim Kulish/Getty Images)
Nvidia envy… Nvidia's AI dominance is casting a shadow over smaller chipmakers. Nvidia’s A100 and H100 AI chips have become the gold standard for lots of large language models (including Open AI’s GPT) and make up 70% of global AI-chip sales. Newer entrants are having a hard time keeping up: so far this year, chip startups have raised just $881M, compared to nearly $2B in the first half of last year. This quarter, the # of US investment deals for chip startups is down by 80% from last year.
AI-dvantage: Nvidia, which has long specialized in graphics chips, reported that its sales doubled last quarter, and said it expects revenue to nearly 3X this quarter.
Feeling the crunch… As Nvidia’s AI power grows, fledgling companies are having trouble meeting investors’ insatiable hunger for market-ready chips. It can cost as much as $500M to advance a single chip design. Some analysts say the surge in AI spend is sapping investment in other types of data-center chips. Intel, which specializes in PC chips, said data-center revenue was down 15% last quarter. But global spending on AI chips is forecast to more than double over the next four years to $106B. Meanwhile:
Being first isn’t always comfy… While Nvidia’s #thriving, it has the burden of supplying most of the industry, and it’s struggling with shortages and production strains. Without lots of AI chip options, and less $$ to help create them, some companies are being forced to delay AI product launches because they can’t get their hands on enough GPUs. But a shift may be coming: chip-making biggie AMD plans to roll out an AI chip to compete with Nvidia’s this year.