(00:00:00) AI Costs Doubling by 2026: The Trillion-Dollar Bill Hitting Your Software Budget
(00:00:39) Enterprise Software Pricing Surge
(00:01:17) Claude Cost Shock And Offshoring
(00:01:48) Failed AI Projects Now Costing Millions
(00:02:18) Codex Windows Control And Biodefense Launch
(00:03:18) Build Focused AI Stacks Now

Enterprise AI costs are no longer a future concern — they're arriving in this quarter's contract renewals. With Amazon, Google, Microsoft, and Meta committing toward $650 billion in AI infrastructure spend in 2026, those costs are being passed downstream into the enterprise software pricing you're already negotiating. Anthropic has moved Claude to usage-based pricing. Oracle, Salesforce, and SAP are raising rates. The era of subsidised AI experimentation is closing, replaced by AI billed like electricity.

This episode breaks down the full cost picture: why some technology leaders are reconsidering offshore engineering as Claude Code bills climb, and what the failed AI initiatives at Pizza Hut and Starbucks — each costing roughly $100 million — tell us about deploying AI without a clear ROI framework.

On the product side, OpenAI expanded its Codex agent to operate Windows applications directly, shifting AI from assistant to operator for development teams. The rollout excludes the EEA, UK, and Switzerland for now — a regulatory signal worth watching. OpenAI also launched Rosalind, a gated biodefense AI system for U.S. government and allied pandemic preparedness teams, introducing a restricted-access deployment model with no prior precedent.

The strategic takeaway: the highest-performing organisations in 2026 won't be sampling fifteen AI tools — they'll be running three to five, deeply integrated into repeatable workflows. Price pressure is accelerating that consolidation now. This is your briefing.

This episode includes AI-generated content.