IBM says AI infrastructure boom is draining software budgets

IBM’s miss is one of the first high-profile confirmations that the AI gold rush is reshaping enterprise IT budgets in a zero-sum way. Customers racing to secure GPUs, memory and networking for AI workloads are pulling spending away from traditional software and some services, at least in the short run. That’s painful for incumbents like IBM, but it also signals that the underlying infrastructure for very large models is being built out aggressively.

For the AGI race, this reallocation of capital is almost certainly accelerative. Billions shifting into high-end servers, specialized accelerators and hardened data-center networks means more capacity for both training and deploying large, agentic systems. At the same time, IBM’s comments about customers prioritizing cybersecurity in response to AI‑enabled hacking and Anthropic’s Mythos model underlines that adversarial pressure is rising in lockstep.

The broader takeaway is that ‘AI’ is no longer an additive line item; it’s crowding out other IT investments. Over time, that could favor platforms that bundle infrastructure, models and application layers into unified stacks, and it may intensify consolidation pressures on mid‑tier software vendors who can’t tie their value proposition directly to AI‑driven outcomes.