AI Summary
Microsoft made a significant $3 trillion bet on its AI tool, Copilot, which is reportedly facing widespread rejection from users despite being integrated into Microsoft 365 and Windows. The company charges $360 annually for Copilot, but internal data suggests extremely low adoption, with only 3.3% of 450 million Microsoft 365 business professionals signing up for the paid service. Users are actively searching for ways to disable or remove Copilot, viewing it as "bloatware on steroids" that consumes resources without providing adequate value. A major point of contention is Microsoft's legal disclaimer, updated in October 2025, stating that Copilot is "for entertainment purposes only" for individual users, undermining its professional utility. Customer satisfaction, measured by Net Promoter Score, plummeted from -3.5 to -24.1 in just two months, indicating strong user dissatisfaction. The "Recall" feature, designed as a photographic memory for PCs, was met with severe privacy backlash due to storing sensitive data in plain text, forcing Microsoft to retool it. Financially, Microsoft is burning through cash, with Capital Expenditure hitting $37.5 billion in the most recent quarter, a 66% jump year-over-year, primarily for building massive data centers and acquiring GPUs for AI infrastructure. Despite this investment, a study by Recon Analytics found only 35.8% of paid Copilot licensees regularly use the tool, compared to 83.1% for ChatGPT. In a desperate move, Microsoft has begun integrating Anthropic's Claude into Copilot, effectively admitting that its OpenAI-based systems are failing critical accuracy tests and diluting its previously touted exclusive OpenAI edge. This multi-model approach is seen as a "Multi-Model Surrender," further confusing users and eroding trust, as 95% of enterprise AI projects reportedly fail to generate profit, with any speed gains offset by time spent correcting AI mistakes. The content concludes that Microsoft's $3 trillion empire is resting on "entertainment only" software, facing a massive consumer rejection that could lead to a significant market correction and strain its partnership with OpenAI.
AI-generated assessment. Verdicts on this page were produced by language models with web search and may contain errors, hallucinations, or out-of-date information. They reflect Bullsift's automated analysis, not editorial judgment. Read the linked sources before relying on any verdict. How this works ·
Claims Extracted (12)
More from The Infographics Show
View all →Trending fact-checks
All claims →- Llama 4 Scout's 10 million token context window degraded sharply past 256,000 tokens, as that was the actual training limit.tech·Seen in 1 video
- Mark Zuckerberg met privately with Alexander Wang, the CEO of Scale AI, for months.tech·Seen in 1 video
- The original Llama model was trained entirely on public data.tech·Seen in 1 video
- Meta released a frontier AI model for free in 2023, allowing developers to download, run, modify, and build businesses on it without API keys, subscriptions, or permissions.tech·Seen in 1 video
- Though they face a 1 in 20 chance of dying on the job and the job is insanely competitive to obtain.tech·Seen in 1 video
- Many people in Los Angeles describe their jobs vaguely, using terms like "I work in entertainment," "I'm in tech," or "I freelance."tech·Seen in 1 video
Want the full picture?
Install the Bullsift Chrome extension to analyze any YouTube video and get real-time fact-checks.
Install Chrome Extension
