IBM CEO Says Open AI Models Are 90% as Good as Frontier Models: ‘Why Pay 10x More?’
Key takeaways
- IBM’s CEO argues that open‑weight AI models offer near‑equal performance at a fraction of the cost, highlighting a cost‑effective shift in enterprise AI strategy and a rebound in its mainframe and software businesses.
- $17.1 billion total revenue
- 9% acceleration in software revenue
- 10% growth in transaction processing
What happened
IBM CEO Arvind Krishna highlighted that open‑weight AI models now cost only 10% to 25% of the token price of frontier models while delivering about 90% of their performance, questioning why enterprises should pay ten times more for the best models. He noted the performance gap has narrowed since late 2024 and that cheaper on‑premises models fit most daily tasks. The comments came alongside IBM’s Q3 2026 earnings, where the company posted $17.1 billion in revenue, a 9% jump in software sales…
Why it matters
The statement suggests a shift in enterprise AI spending toward more affordable, sovereign‑AI solutions, potentially reducing capital and operating costs for large organizations. It also signals that IBM’s mainframe and software businesses are recovering, with software revenue accelerating and transaction processing…
Short-term and longer-term
Short-term. Over the next few days, watch IBM’s stock for reactions to the AI cost narrative and the Q3 earnings beat.
Longer-term. For the coming months, IBM’s focus on affordable, sovereign AI and the strengthening of its mainframe software could improve its long‑term earnings profile.