Archive

Today's signal

Most enterprises now depend on AI in production, but only about a third have an operating model to own it, and that gap is now the technology leader's job.

In the enterprise

Two in three enterprises now depend on AI, but few have an operating model for it

Source Humatica via Business Wire

What happened

66% of enterprises in the study run a process or product that would degrade or break if AI were switched off, but only about 35% have a formal operating model for AI. Leaders named their own organisation as the constraint 2.4 times as often as the technology, and 42% of production AI use cases started in the business, not in IT.

So what

Building AI has become easy; owning it, meaning integrating, securing, monitoring and paying for it, is now the executive job. Shared guardrails, cost visibility and clear ownership are what turn scattered AI use into an operating model.

Talent & organisation

Australian companies face "AI washing" claims over job cuts and offshoring

Source Reuters via CNA

What happened

Qantas is cutting 850 jobs and outsourcing some corporate functions in an AI-driven overhaul with Accenture, and unions accuse it of using AI as cover for offshoring. The article notes that Atlassian and WiseTech also attributed job losses to AI this year.

So what

Being honest about AI's real impact on the workforce is becoming a credibility test for technology leaders. Separating what AI actually automated from what was simply a cost decision builds trust inside and outside the company.

Cost, cloud & markets

Nvidia's change of tone on financing hits CoreWeave and the other neoclouds

Source CNBC

What happened

After Nvidia's earnings, analysts spotted new footnote wording: its commitments to neoclouds apply only if those providers secure financing, and data-centre delays can cancel them. CoreWeave and Iren fell about 3%, while separate Grayscale analysis shows H100 rental rates up roughly 50% over the year.

So what

AI compute is now a supply-chain and vendor-risk question, not just a line item. Boards will increasingly ask where your capacity comes from, what it costs per outcome and what happens if a provider stumbles.

The idea

Jack Dorsey on rebuilding Block "around intelligence"

Source HBR IdeaCast

What happened

Dorsey explains that Block's 2024 move from business units to disciplines was about shared standards and tools, and that this made the current AI-driven reorganisation possible, layoffs included. His approach is to gather opposing views first, accept losing some credibility in the short term, and keep explaining why the change matters until the results prove it.

So what

People judge a reorganisation less by the org chart than by how leaders take them through it. Seek out dissent early, explain the why repeatedly, and set a short horizon to prove the decision.