{
  "id": 3682720,
  "title": "Michael Burry Says This $21 Billion AI Startup Could Crack Nvidia’s Empire",
  "url": "https://urgent.news/2026/08/25/michael-burry-says-this-21-billion-ai-startup-could-crack-nvidias",
  "topic": "ai",
  "section": "AI",
  "published": "2026-08-25T19:03:32.000Z",
  "source": {
    "name": "Yahoo Finance",
    "slug": "yahoo-finance",
    "url": "https://finance.yahoo.com/technology/ai/articles/michael-burry-says-21-billion-190332718.html"
  },
  "original_language": "en",
  "account": "Michael Burry, known for his role in the 2000s financial crisis with \"The Big Short,\" has identified a potential competitor to Nvidia, the dominant AI chipmaker. Burry believes that Etched, a four-year-old AI chip startup, could pose serious competition to Nvidia. Etched recently secured a $700 million funding round led by Jane Street, valuing the company at $21 billion, more than double its previous valuation. The startup focuses on AI inference, where performance per token, latency, and power efficiency are critical as AI models move from training to serving at scale. Burry points to an industry source suggesting Etched could deliver ten times the performance at lower cost compared to Nvidia.\n\nThe talent angle for Nvidia is also concerning: around 15% of Etched's 400-plus employees previously worked at Nvidia. Etched's leadership includes Brian Loiler, a former 22-year Nvidia employee who helped develop HGX and DGX systems. If Etched's economics hold up as it scales, taking a meaningful share in inference could pressure Nvidia's pricing power and margins. In fact, Nvidia's own actions suggest they recognize the importance of specialized inference hardware. Last year, Nvidia partnered with Groq, hiring key members of its team, including founder Jonathan Ross, and integrating Groq technology into its own products.\n\nHowever, Etched still needs to demonstrate its ability to manufacture and deploy systems at Nvidia's scale. Its performance claims have not been independently verified across the wide range of workloads Nvidia supports. Nvidia's ecosystem, which includes software like CUDA, networking, rack-scale systems, and more, creates switching costs that a faster specialized chip alone cannot easily overcome. As of the latest quarter, Nvidia reported an impressive $81.6 billion in revenue, up 85% year over year, with $75.2 billion coming from data centers. Analysts maintain a strong outlook on Nvidia, with BofA suggesting a $350 target, implying more than 50% upside from current levels. Institutional investors, such as Fisher Asset Management and AQR, continue to hold significant positions in Nvidia, with no signs of panic.\n\nWhile Etched represents a credible threat, it is not yet an \"Nvidia killer.\" The bigger risk is that specialized inference chips could gradually erode Nvidia's dominance and compress its margins. Despite Nvidia's continued rapid growth and relatively low forward earnings multiples, the stock appears attractive if there's meaningful weakness that creates a better entry point. However, Burry has highlighted a competitive threat that Nvidia investors cannot ignore. Despite acknowledging Nvidia's potential as an investment, Burry believes certain AI stocks offer greater upside potential with less downside risk. For investors seeking an extremely undervalued AI stock that could benefit from Trump-era tariffs and the onshoring trend, Burry suggests exploring another free report on the best short-term AI stock.",
  "summary": null,
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}