Everyone thinks Google is losing the AI race. I think the more interesting question is whether Microsoft is quietly becoming the next Nokia.
Why This Matters and What Nokia Actually Teaches Us
The Nokia comparison gets thrown around a lot. But the lesson is usually misread. Nokia didn't fail because they built bad hardware. They failed because they kept optimising for the wrong layer. They were excellent at phones as physical objects. They missed that the phone was becoming a platform, and that the platform layer would own everything below it.
IBM made the same mistake a decade earlier. They thought Windows was just software sitting on top of their hardware. They let Microsoft license it freely because they couldn't imagine the software layer eating the hardware layer's margin. Then it did.
So the pattern isn't really about one company being slow or arrogant. It's about companies that are dominant in one layer failing to see when a new layer above them becomes the one that actually captures value.
That's the frame I'm applying to Microsoft right now.
What Is the Intent Layer, and Why Does It Decide Who Wins?
The intent layer is the interface where a user expresses what they want, and a system figures out how to fulfil it. Google built the most valuable intent layer in history. You type something into a search bar, and that act of typing is worth billions in advertising revenue because it reveals commercial intent with almost no friction.
AI is rewriting that layer from scratch. Instead of typing keywords, you describe what you need. Instead of scanning ten blue links, you get an answer. The question is: who owns that moment when you express your intent?
This is the game being played right now. And I think it's the only game that matters for the next decade.
My current read is that Microsoft is betting heavily on the wrong side of it.
What Most People Get Wrong: The OpenAI Investment Isn't Proof Microsoft Is Winning
When Microsoft put serious money into OpenAI and shipped Copilot across its entire product suite, most people called it a masterstroke. The narrative was clean: Microsoft had been sleeping, then woke up, and now they're back in the race.
I think the more complicated version is more likely true.
Microsoft's core identity is still Windows, Office, Azure. Those are the gravity wells the whole business orbits around. Copilot is being built on top of those layers, which means it's an enhancement to existing products rather than a new intent surface in its own right. That's a fundamentally different strategic posture than building the thing people come to first.
The OpenAI investment, if anything, confirms to me that the smartest people inside Microsoft's shareholder base sensed the same risk I'm describing. You don't make a bet that size unless you're aware that your core business might not be the right layer to compete from.
My ChatGPT sparring partner pushed back on this. It argued that Microsoft's infrastructure position, through Azure and its data centre footprint, gives it a durable role even if it loses the intent layer battle. And that's a fair point. It's also the more responsible take. I have 13 subscribers and no fiduciary duty. The AI has a slightly heavier responsibility to be balanced.
But "durable infrastructure role" is exactly what IBM had. And it didn't protect them from a long, slow margin erosion while the layer above them captured all the exciting value creation.
What Google Actually Gets Right (And Why Heritage Matters)
When Gemini launched, the narrative was that Google had panicked and rushed something out. Maybe. But Google's situation is more interesting than that.
They have been working on transformer-based AI internally for years. The original Transformer paper came from Google Brain. They were running large-scale AI infrastructure before OpenAI existed as a company. That's not a talking point. That's a real technical and institutional inheritance.
As Mourinho would say: this is heritage. And heritage is important.
More concretely, Google still owns the intent layer. Every day, billions of people go to Google to express what they want. That's the position Microsoft doesn't have and can't easily buy. Bing has been a capable product for years. It has never meaningfully changed that behaviour.
Gemini plugged directly into that existing intent surface. Copilot is still trying to convince people to open a new window.
The Windows Question
Here is the hot take I'll commit to: Windows will lose in the intent layer game.
Not immediately. Not catastrophically. But slowly, in the way that matters: fewer young users forming habits around it, more workflows that start in a browser or a voice interface or a mobile app, more developers building for platforms that aren't Windows-first.
The one real escape route is the OpenAI relationship. How much influence can Microsoft accumulate or convert before OpenAI becomes something it can't fully control? That's the question I'll be watching closely over the next twelve months. It will tell us a lot about whether this is a Nokia story or something with a different ending.
I don't have the answer yet. I'm not sure anyone does. But the question itself seems worth sitting with.
Nokia thought the phone was hardware. IBM thought Windows was just software. Microsoft might be making the same mistake with AI.
What to Actually Do
- Watch where new intent habits form, not who has the biggest AI budget. Budget follows users, not the other way around.
- Pay attention to how OpenAI evolves its direct relationship with end users. Every step toward consumer independence is a step away from Microsoft's leverage.
- Track whether Copilot gets used as a starting point or a finishing tool. The distinction is everything.
- Notice which layer developers are building on. Developer behaviour predicts consumer behaviour by about two years.
- Keep an eye on Google's Gemini integration inside Search specifically, not the standalone app. The intent layer reinforcement is happening there.
- Treat the Microsoft-OpenAI relationship as the most interesting corporate drama in tech right now. The details will matter.
The most consequential battles in tech are rarely about who has the best technology. They're about who owns the moment when someone decides what they want next.