In early 2025, when Microsoft proudly announced that 30% of its source code was written with AI, many viewed it as a big leap forward for the software industry.
But a few months later, reality showed a completely opposite picture. A series of Windows 11 updates became a disaster: the October update left Windows recovery Environment unable to receive the keyboard and errors, the August version caused hard drive data loss, and the larger version 24H2 continuously had installation errors.
From un loc locomotable websites, frozen apps to Windows not understanding the Up date and Shut Down command, users feel like they are the rates of incomplete AI testing.
Meanwhile, the uncertain atmosphere about AI is becoming increasingly clear at the Cerebral Valley Conference 2025.
Investors and founders were asked to choose the $1.5 billion AI company they thought the most likely to fail.
The results were shocking as Perplexity took the lead, surpassing OpenAI; the Cursor, figure, Harvey, Mercer, Mistral and Thinking Machines groups shared the next positions.
An unofficial survey but accurately reflecting market sentiment is that the AI bubble may be taking a break.
Perplexity currently processes 780 million queries per month, a large number but still just like a dust bin when placed next to Google with 13.7 billion queries per day.
The search game, which was already at a standstill even when Microsoft integrated ChatGPT into Bing but still only increased its market share to 4.31%, shows that this is a money-spending battle, not a profit-making one.
Perplexity is burning capital, and is caught up in a lawsuit with Amazon, causing many investors to question whether the AI search model can survive financially.
Amid that wave of anxiety, Microsoft CEO Satya Nadella posted a warning about the partnership exploited by AI.
Mr. Satya Nadella emphasized the importance of maintaining technology sovereignty and said that the ideal partner must promote both sides to move forward, as in the example between Microsoft and OpenAI.
However, observers say he has ignored the reality: users are not concerned about business ethics, they only see Windows 11 constantly stuffing AI into their experience, but it does not bring clear benefits.
Not only Microsoft, another name also shows the struggles of the industry: Rabbit. The startup was once famous for its AI R1 device, which was rumored to be behind staff salaries, but at the same time still promoted a new AI hardware product.
Many analysts see this as evidence of the "we can't pay, but let's believe in vision" period, a dangerous signal of the AI winter.
Amid that chaos, Mozilla chose a different approach. Instead of AI everywhere like Microsoft or specialized AI browsers like Google, Firefox launches AI window, a web browser mode with an integrated chatbot on the sidelines, which only works when users are proactive.
Mozilla affirms that users are completely free to decide whether to use AI or not, which model to use and when to use.
This is a rare philosophy of AI respects users in the race that is full of pressure in todays data economy.
At the same time, Windows leader Pavan Davuluri admitted that user feedback is focusing on three factors: reliability, performance and simplicity, which are things Windows 11 is gradually losing.
AI features are not yet perfect, the forced changes to the interface and use of clouds have led many to publicly criticize Microsoft for living in an AI bubble.
They want stability like Windows 7, not an operating system that turns into an AI technology billboard.
Although AI can be the future, the lesson in 2025 shows that the future cannot be built by haste, imposition or illusion. Users need real value, not promises.
