Today's signal
AI is speeding up hard platform work at the same time as trillions of data-centre commitments sit off the big tech balance sheets, so cost, risk and organisation now belong in the same conversation.
In the enterprise
ANZ moves a 1.7PB data platform to AWS in six months, using AI to speed it up
Source iTnews
What happened
ANZ moved its in-house Cloudera Hadoop data platform onto Amazon's cloud in six months, against an 18-month plan. The move covered 300 million files and 142,000 tables. The bank says AI agents cut the migration effort by about 35% and halved defects and rework.
So what
This is the kind of story boards want to hear: AI used to speed up hard platform work and reduce risk, not just added as a chatbot. The useful version always includes the scale of the work, the time saved and the effect on quality.
Security & risk
Companies are using AI agents faster than they can explain the agents' decisions
Source IDC, commissioned by Quantexa
What happened
In a survey of 366 technology leaders at banks, insurers and government agencies, AI agents already run 21.6% of the processes they could handle, and leaders expect 32.1% within a year. Only 8.7% have a reliable record of why their agents make the decisions they do.
So what
The problem now is oversight and explaining an agent's decisions, not how good the model is. The study was paid for by a vendor, so treat the exact figures with care, but the trend is the point.
Cost, cloud & markets
The Fed finds more than US$3 trillion of AI data-centre commitments that aren't on Big Tech's balance sheets
Source Fortune
What happened
The US Federal Reserve's November Financial Stability Report has a new section on debt behind data centres. It counts more than US$1T in signed leases that haven't started yet. It also counts more than US$2T in contracts and guarantees that commit Amazon, Alphabet, Broadcom, Meta, Microsoft, Nvidia and Oracle to pay whether or not they use the capacity. Fed Governor Barr said the seven companies' recorded debt is expected to pass US$800B by the end of the year.
So what
The cost of computing power is now something CFOs and financial regulators worry about. Technology leaders need to be able to join that conversation, not only the architecture one.
The idea
To get value from AI, redesign the organisation, not just how individual tasks are done
Source McKinsey Leading Off
What happened
McKinsey argues that as companies use more AI, "the existing organization can become the constraint." It names four connected shifts, starting with organising work around outcomes instead of functions and hierarchy. Its 2026 survey found 80% of people say AI has made them more productive, but far fewer see it lifting company earnings.
So what
As companies use more AI, the existing organisation can become the constraint. The work is to redesign around outcomes, not just to speed up individual tasks, and to put clear authority behind that change.