Particle.news

Download on the App Store

AI Buildout Accelerates as Citi Sees $2.8 Trillion Capex and Nvidia Sets Record Valuation

Rising debt funding alongside vendor-style deals is drawing bubble warnings from prominent investors.

Overview

  • Citigroup raised its forecast for AI infrastructure investment to more than $2.8 trillion through 2029 and now projects hyperscalers will spend $490 billion by the end of 2026, with each gigawatt of compute estimated to cost about $50 billion.
  • Nvidia and OpenAI formalized a plan to deploy about 10 gigawatts of Nvidia systems over roughly five years, with Nvidia committing up to $100 billion progressively, and the first gigawatt expected in the second half of 2026 on the Vera Rubin platform.
  • The financing mix is shifting toward borrowing, with Oracle issuing $18 billion in bonds as it scales cloud capacity for multiyear AI contracts, and OpenAI reportedly discussing leasing Nvidia chips to lower upfront costs.
  • Nvidia shares hit a 52-week high and a $4.5 trillion market capitalization, as analysts raised targets including Citi to $210 and Jefferies to $220, citing sustained AI demand.
  • Bain estimates AI providers could face an $800 billion annual revenue shortfall by 2030 relative to compute needs, while figures like James Anderson and Steve Hanke caution that rapid valuation jumps and circular deal structures echo past bubble dynamics.