Most people are reading the Amazon and Anthropic deal as a funding story but it clearly isn’t. It’s an infrastructure story.
$25 billion into Anthropic, with Anthropic committing over $100 billion back into AWS over the next decade. That’s not a financial transaction – it’s two organisations structurally aligning around who controls the layer that enterprise AI runs on. And that layer is what actually matters.
Obsessing over which AI model is best is becoming a distraction faster than most people realise. Models are commoditising, but infrastructure is where true leverage is being built, quietly and deliberately.
What that means in practice? Once your data pipelines sit on a platform, your integrations follow, your cost base follows and your institutional knowledge of how to operate on that stack follows. Switching later isn’t just expensive – it’s the kind of decision you end up regretting for years.
Most UK enterprise organisations I’m speaking to are still thinking at ‘tool’ level: which AI writes better emails and which one summarises documents faster. That’s not necessarily the wrong conversation to be having, but it’s definitely not the most strategic.
The strategic question is simpler and harder at the same time: where does your AI stack live and who controls it?
The next 12 to 18 months will lock a lot of this in. The organisations that treat that as a deliberate decision will be in a very different position to those that arrive at it by accident.
Leave a Reply