Microsoft $MSFT has thrown in the towel in the AI chatbot race

Microsoft $MSFT has thrown in the towel in the AI chatbot race. Copilot is losing billions and the company is cutting it back...

The company that helped bring ChatGPT to the whole world has admitted it won't win the battle for a personal AI assistant. $MSFT shares reacted to the news on Friday by rising 4%. It's a bit of a paradox, because the whole year has been about getting AI into as many devices, programs and apps as possible. Yet it all makes sense.

Two years ago, Microsoft split Copilot development into two teams. One built an assistant for work, the other for "regular" users. Microsoft's plan was clear: since we helped spread OpenAI, we can do it with our own product too. They couldn't.

ChatGPT quickly passed the 1 billion user mark, Google $GOOG deployed Gemini across the entire Android ecosystem, and relatively few people used Microsoft's personal chatbot offering.

Microsoft was thus forced into a radical step. It is merging the consumer and work versions of Copilot into a single product aimed at businesses and leaving the personal chatbot market to OpenAI, Google and now also Meta $META with Muse.

The newly created app focuses on 3 tools:

- a chat hub

- programming tools

- an agent named Autopilot

The company also changed subscription fees. Copilot now starts at $30 per user per month and Microsoft is adding consumption-based billing to per-user licenses. For the businesses Microsoft is now mainly focusing on, this makes sense. It couldn't afford that with regular users.

Microsoft 365 Copilot has over 30 million paid licenses, while in April there were just over 20 million. GitHub Copilot is used by 50 million developers. This is where Microsoft makes money. A chatbot for the masses, by contrast, was an expensive chase for users who already have ChatGPT or Gemini on their phone.

Shares strengthened by 4% on Friday. Oppenheimer raised its target price to $570 from $515. Some analysts called the new app Microsoft's long-speculated "super app."

But this strategy also has a downside.

- Usage-based billing means less predictable revenue, because part of the income will be tied to how often customers actually launch the agent tools.

- The second thing is investment. Capex for the fiscal year reached almost $116 billion, 80% more than the previous year, and for calendar 2026 the company expects $190 billion. After adjustments from the latest results, that works out to $175 billion, with $MSFT only extending the depreciation period, so the total amount remains unchanged but is spread over a longer time.

Do you agree that Microsoft did the right thing by leaving consumers to $GOOG, $META and OpenAI? I think it was a good move.

In my opinion, a completely logical step. Microsoft was historically a B2B and B2C company, but in recent years it has transformed and earns primarily from B2B (the corporate part such as M365, Azure, and Dynamics accounts for about 80% of revenue, consumer for the remaining about 20%). So why should it burn billions on end users who have ChatGPT or Gemini on their phones for free? Moreover, there is a fundamental difference: Microsoft funds AI from what companies pay for licenses, whereas Meta and Google primarily drive it from advertising. For them, every user is additional ad space, for Microsoft just an extra cost. At first glance, everyone does "the same thing," but business-wise it is far from the same.

US stocks drop as oil prices and bond yields spike