Nvidia is planning a price increase of approximately seventeen per cent on its flagship Blackwell-family AI processors, expected to take effect in the fourth quarter of 2026. The increase adds a complicating variable to enterprise AI budget models that have, over the past twelve months, been revised primarily downward on the basis of falling inference costs. Training and fine-tuning workloads — which remain GPU-compute-intensive even as inference efficiency has improved — are likely to see the most direct impact. Enterprises planning significant model customisation programmes in 2027 should incorporate revised hardware cost assumptions into their business cases. The price increase also has implications for competitive dynamics in the inference infrastructure market: at higher chip costs, the economic case for custom silicon — such as the Jalapeño chip OpenAI unveiled this week — improves relative to purchasing Nvidia hardware at the margin, accelerating the vertical integration trend already visible across the major laboratories.
Venture
Instinct Raises $1 Billion at $10 Billion Valuation — Four Times Its Valuation From One Month Earlier
Series C led by Sequoia, Benchmark, and Coatue (28 September). One month prior: $2.5B valuation. Wha…