Summary of Key Points
Microsoft's 2026 Build conference was shortened from four days to two, with half the number of attendees, and the location was moved from Seattle to San Francisco, signaling a trend of "consolidation." This move reflects multiple pressures Microsoft is facing: a decline in its stock price (a market value reduction of $440 billion), low usage rates for Copilot (only 5 out of 100 users are actually using it), and a loosening of cooperation with OpenAI (which has turned to AWS). At the conference, Microsoft unveiled seven self-developed AI models (the MAI series) and collaborated with NVIDIA to launch the RTX Spark chip in an attempt to "reinvent the PC." However, Microsoft's previous attempts at "reinvention" (with the Surface RT and Nokia acquisition) ended in failure. Whether there is a real demand for these AI agents and whether they will be successful remains uncertain.
1. The Reduction of the Build Conference: What Lies Behind Microsoft's Tightening Up?
The changes to the Build conference are quite significant: the duration was cut from four days to two, the number of attendees decreased from 5,000 to 2,500 (with application and approval required), and the location was changed from its home base in Seattle to San Francisco, a hub for AI technologies. The new slogan clearly emphasizes a shift towards practicality. Such changes are not incidental; when a company reduces its most important developer conference, it is usually signaling an internal reevaluation of its strategies.
Why the reevaluation? Microsoft is facing tough times: its stock price has dropped by a quarter from its historical high, and its shares plummeted by 10% in January this year following its financial report (resulting in a loss of $440 billion). Wall Street is questioning the value of spending $19 billion annually on AI infrastructure. The reduction in the conference size reflects a desire to focus on practical results rather than empty rhetoric, especially considering the poor actual usage rates of Copilot.
2. The Embarrassing Situation with Copilot
Copilot was Microsoft's major AI highlight over the past two years, with Satya Nadella claiming it would transform how all knowledge workers work. However, reality has been harsh: out of the 400 million corporate users of Microsoft 365, only 20 million have subscribed to Copilot, and only about 5 people out of 100 use it regularly. Ironically, Copilot experienced a major outage the day before the conference, leaving companies that rely on it without access to the service.
This highlights the challenges in bringing AI products to market. The technology may sound promising, but users often find them either unusable or unnecessary. Wall Street reacted negatively, with Microsoft's stock price falling by 3.6% on the day of the conference, as investors wanted to see tangible results rather than empty promises.
3. The Strained Relationship with OpenAI
Previously, OpenAI was a key partner in Microsoft's AI efforts, with Copilot using its GPT technology. But at this conference, Microsoft highlighted its own MAI series, emphasizing that they were trained from scratch without any external AI input. This change indicates a shift in strategy: in April of this year, Microsoft and OpenAI revised their partnership agreement, allowing OpenAI's models to be used on AWS without further payment. To make matters worse, OpenAI has since signed a large deal with Microsoft's cloud computing competitor, AWS. Microsoft is concerned about being left behind; after IBM supported it, it was surpassed by Microsoft, and now it fears the same could happen with OpenAI. By launching its own AI models, Microsoft signals to the market that it can continue to thrive even without OpenAI—its GitHub Copilot will soon switch to its own Polaris model in August.
4. Collaborating with NVIDIA to "Reinvent the PC": Will This Lead to the Same Mistakes as the Surface RT?
At the conference, Jensen Huang unveiled the RTX Spark chip and announced an attempt to "reinvent the PC," similar to how smartphones transformed mobile devices. However, Microsoft's previous attempts at reinvention failed: the Surface RT was a disappointment, with 6 million units left unsold and resulting in a $900 million write-down; the Nokia acquisition also ended in a $7.6 billion loss and the dismissal of 7,800 employees.
What makes this effort different? There is a impressive lineup of manufacturers (including ASUS and Dell) working on the new devices, and NVIDIA has provided a three-generation chip roadmap. However, there are challenges: Apple's M-series chips dominate the high-end Arm notebook market, and there is a global shortage of AI-related memory. More importantly, do users really want an AI system that makes decisions for them? Just because the technology is feasible does not mean users will embrace it.
5. The Uncertainty Around the Demand for AI Agents
Satya Nadella presented AI agents as the next major opportunity, describing them as AI colleagues that can work independently and make decisions on behalf of users. However, the poor adoption of even basic AI assistants suggests that jumping straight to more advanced concepts may be premature.
Nadella himself is not entirely confident about this approach; he did not deliver any definitive statements at the end of the conference, instead discussing the potential for AI to centralize power and reduce human autonomy, indicating that he is aware of the risks. The success of Azure was based on existing demand (the trend towards cloud computing), but the demand for AI agents has not yet been proven. Whether users want AI assistants or "bosses" remains uncertain, and Microsoft's future depends on whether this need actually exists.
Conclusion: Can Microsoft's "Self-Saving Strategy" Succeed?
Microsoft is in a position of being questioned by the market: it needs to reduce its dependence on OpenAI and launch new hardware (in collaboration with NVIDIA). However, past failures and current data warn that success is uncertain. The future of AI agents hinges on whether users truly need such systems. If there is no real demand, even the most ambitious efforts could end in disappointment, similar to the fate of the Surface RT. Whether Nadella's strategy will work depends on how the market responds.