Microsoft Copilot Surpasses 20 Million Paid Users, Google Cloud Hits $20 Billion Quarter for the First Time, OpenAI Announces Images 2.0

Microsoft Copilot Surpasses 20 Million Paid Users — "Nobody Uses It" Is a Thing of the Past

In its earnings report, Microsoft announced that its AI assistant Copilot has surpassed 20 million paid users.

The industry long held the view that "nobody uses Copilot," but Microsoft emphasizes that engagement, not just user counts, has been steadily growing. Deeper integration into the Office product suite, combined with more robust enterprise security features, appears to have lowered the barrier to adoption.

Skepticism about the actual usage rates of AI tools has persisted, but these numbers suggest that Copilot is at least starting to gain traction in enterprise use. That said, it remains unclear what share 20 million represents of Microsoft 365's overall user base (several hundred million), and how retention from free trials is trending.

Google Cloud Tops $20 Billion in Quarterly Revenue for the First Time — But Says It "Could Have Grown More"

Google Cloud has surpassed the $20 billion mark in quarterly revenue for the first time. The milestone was driven by surging AI demand, but Google has revealed that infrastructure capacity constraints meant it "could have grown even more."

Demand for compute resources such as GPUs and TPUs needed for AI workloads is outstripping supply, underscoring how major cloud providers—not just Google—are racing to expand their infrastructure. Behind the AI boom, the physical constraints of data centers are becoming a bottleneck for business growth.

OpenAI Announces Latest Image Generation Model "Images 2.0" — Improved Accuracy with Thinking Mode

OpenAI has announced the latest version of its image generation model, Images 2.0.

The headline feature is a Thinking mode that interprets instructions more deeply. Conventional image generation sometimes failed to accurately capture the intent of complex prompts, but Thinking mode analyzes and structures the prompt before generation, aiming to produce output that stays closer to the user's intent.

The model is also expected to find use in practical applications such as building presentation decks — a sign that image generation AI is entering the stage of transitioning from a "toy" to a "tool for work."

AlphaGo's Creator Warns That "AI Is on the Wrong Track"

David Silver of DeepMind, who led the development of AlphaGo, has made remarks questioning the current direction of AI research.

According to a WIRED interview, Silver advocates the concept of "ineffable intelligence" and argues for the importance of approaches in which AI acquires knowledge in the form of tacit understanding that cannot be put into words. Drawing on AlphaGo's experience of discovering strategies beyond human comprehension through reinforcement learning, he contends that the currently dominant LLM-centric approach alone is not enough.

Combining reinforcement learning with LLMs is already an active area for many researchers, but a direct critique from the person who drove AlphaGo's success could send ripples through the industry.

U.S. Accuses China of "Industrial-Scale AI Model Distillation" — A New Chapter in the IP Debate

The U.S. government has accused China of carrying out AI model distillation at an industrial scale.

Model distillation is a technique for training one model using the outputs of a high-performance model. The U.S. government contends that Chinese AI companies are making heavy use of U.S.-built models' APIs and using the outputs as training data to build their own models.

This issue represents a new type of intellectual property risk that traditional copyright and patent frameworks cannot fully address, and it could affect the entire AI industry. A growing number of companies are prohibiting distillation in their models' terms of service, and further legal developments are expected.

Shopify's "AI-First" Engineering Practices Go Public

Venture capital firm BVP has published a detailed look at Shopify's AI-first engineering practices.

Shopify has adopted AI-assumed development flows across its entire engineering organization, and what sets it apart is that AI tools are positioned not as auxiliary aids but as core development tools. The write-up covers concrete practices, team structures, and the design of evaluation metrics, making it a valuable reference for other companies pushing forward with AI adoption.