IBM CEO Arvind Krishna says there is no AI bubble after all

December 1, 2025·1 hr 10 min·5 clips

IBM's CEO says pushing Watson into healthcare too early was inappropriate—here's why they got the timing wrong.

Nilay Patel frames the interview around the puzzle of what IBM actually does in the current era, noting that Watson's 2011 Jeopardy victory generated massive brand awareness but the subsequent decade of IBM Watson healthcare initiatives produced few visible consumer results. Arvind Krishna concedes directly that pushing Watson into healthcare was inappropriate and that the monolithic, non-modular architecture of the original Watson was the wrong product approach even if the underlying technology was directionally correct. He traces the lineage from Watson's statistical and neural network models to modern deep learning to LLMs, explaining that the critical unlock of LLMs was eliminating the need for human-labeled training data for every new task, which collapsed costs by 100x and made AI deployable at industrial scale across many different enterprise use cases simultaneously. Krishna is explicit that IBM has no consumer AI product and is not competing with ChatGPT or Gemini at the consumer layer, arguing that the enterprise layer is where durable value will be created because it requires trust, security, hybrid cloud integration, and workflow embedding that consumer AI products cannot provide. On the bubble question, Krishna draws the dot-com parallel: the infrastructure and applications built in the late 1990s were eventually realized, but the transition took longer and caused more disruption than anticipated, including significant negative consequences for workers and communities. He argues the AI infrastructure and model investments of today will similarly prove correct over a longer arc even if near-term hype exceeds near-term reality. The conversation covers IBM's mainframe business, which remains substantial and cash-generative, its Red Hat open-source acquisition as a hybrid cloud foundation, and the decision to invest heavily in quantum computing as a multi-decade bet on a platform shift that could render current cryptography and optimization approaches obsolete. Krishna is notably candid throughout, acknowledging mistakes and offering specific technical explanations rather than marketing language.

As heard by us

Arvind Krishna argues that AI's biggest opportunity is less app-store rerun than enterprise and physical-economy coordination.

Arvind Krishna's case for AI avoids the usual consumer-app victory lap. When Nilay Patel compares the boom to investors trying to rerun the smartphone playbook, Krishna accepts that phones changed the front end of the economy, then points to everything still sitting behind it:…

Read the full review in PlayNext →

Why you'd press play

Arvind Krishna walks into a bubble question and refuses to hand back an easy answer.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from Decoder with Nilay Patel