Databricks CEO: Open-Source Models Drive Compute Demand
GavinSBaker · x · 2026-07-20
Databricks CEO Ali Ghodsi shared insights on open-source models and compute demand:
- Model Usage Strategy: Enterprises prefer using expensive frontier closed-source models only for the most complex problems, delegating routine tasks (like extracting and summarizing business data) to cheaper open-source models.
- Open-Source Demand Surge: Databricks hosts open-source models like Kimi, but the overwhelming demand has triggered GPU shortages across multiple regions, nearly exhausting capacity in Asia.
- Compute Expansion: Compute demand continues to surge in Japan, South Korea, the US, and India. This is the core reason behind the company's recent massive fundraising—to buy more GPUs and bridge the gap.
Comments point out that cheaper open-source model tokens are actually driving the incremental growth in overall compute demand.
Related event: Databricks CEO Warns of GPU Shortage Amid Open-Source AI Boom(2 posts)→
More from Venture
- ChatGPT sends just 0.62% of Google's traffic — but it's a branding channel, argues iPullRank — VeryWellVersed · 2026-10-26
- Ant's $46.4B annualized enterprise software spend already equals ~3.2% of global software spending — BenBajarin · 2026-10-26
- Which VCs stand to make blockbuster gains from Anthropic's eventual IPO — nmasc_ · 2026-10-26
- citrini pushes back: '80B losses vs 16B revenue' confuses net income with burn — citrini · 2026-10-26
- Anthropic files S-1: IPO expected between Nov 2026 and Feb 2027, valuation up 15x — alliekmiller · 2026-10-26
- Anthropic S-1 reveals 48 cents of compute cost per dollar of Q2 revenue — nmasc_ · 2026-10-26