PLUS: A $45B bet on Anthropic and the first 100 billion-star galaxy simulation
Good morning
Google's newest and most powerful model, Gemini 3, has arrived with a major new capability: generating interactive software on demand. The model is being integrated directly into its core products, including Search, from the very start.
This shift blurs the line between asking a question and using a dedicated application. What does it mean for the future of software when an AI can build a functional tool right inside your search results?
In today’s Next in AI:
Google's software-generating Gemini 3
Microsoft and NVIDIA's $45B Anthropic deal
AI simulates the 100B-star Milky Way
Inside Microsoft and NVIDIA's AI superfactories
Google's Gemini 3 generates interactive software on the fly
Next in AI: Google has launched Gemini 3, its most powerful AI model yet, integrating it directly into Search, the Gemini app, and developer platforms from day one. The new model features state-of-the-art reasoning and can generate dynamic, interactive user interfaces on demand.
Decoded:
Gemini 3 introduces new generative UI experiences in Search, allowing it to create interactive tools like a loan calculator or a physics simulation directly within the results.
For developers, Google released Antigravity, an agent-first platform, and announced that Gemini 3 Pro is now rolling out in GitHub Copilot to help build and debug code.
The model sets new performance records across multiple benchmarks, taking the top spot on the LMArena Leaderboard and demonstrating PhD-level reasoning on complex exams.
Why It Matters: By integrating its most advanced model into core products at launch, Google is moving with newfound speed and confidence. The ability to generate functional software on demand signals a significant shift, blurring the line between asking a question and using a dedicated application.
AI's New Power Alliance
Next in AI: Microsoft, NVIDIA, and Anthropic announced a massive partnership that brings Anthropic's Claude models to Azure. The deal includes Anthropic's commitment to purchase $30 billion in Azure compute, backed by a combined investment of up to $15 billion from Microsoft and NVIDIA.
Decoded:
This move helps Microsoft significantly diversify its AI bet beyond OpenAI, giving Azure enterprise customers more choice in frontier models and reducing its reliance on a single partner.
The deal establishes a deep technical collaboration, where Anthropic and NVIDIA will work together to optimize Claude models for performance and efficiency on future NVIDIA chips like the Grace Blackwell and Vera Rubin systems.
Despite the major Azure commitment, Anthropic confirmed that Amazon remains its primary cloud provider and training partner, highlighting the increasingly common multi-cloud strategy for leading AI labs.
Why It Matters: This alliance creates a powerful, vertically-integrated stack combining top-tier models, cloud infrastructure, and the underlying hardware. For developers and businesses, this intensified competition will likely accelerate innovation and expand access to powerful new AI capabilities.
AI Models the Milky Way
Next in AI: Researchers used a new hybrid AI to create the first-ever simulation of the Milky Way tracking over 100 billion individual stars. The method overcomes a massive computational barrier, running hundreds of times faster than previous models.
Decoded:
Before this, simulations could not track individual stars at a galactic scale, instead grouping them into particles and losing crucial detail on small-scale events like supernovae.
The new approach uses a deep learning model to predict the after-effects of these events, allowing the main physics simulation to run much more efficiently without getting bogged down.
This hybrid technique is hundreds of times faster, reducing simulations that once took around 315 hours to a fraction of the time.
Why It Matters: This achievement showcases how AI can serve as a powerful accelerator for scientific discovery, moving beyond simple pattern recognition. The technique provides a new blueprint for modeling other complex systems, from climate change to oceanography.
Inside the AI Superfactories
Next in AI: NVIDIA and Microsoft are deepening their partnership to build next-generation 'AI Superfactories,' integrating NVIDIA’s latest Blackwell GPUs and Spectrum-X networking tech directly into Azure's AI superfactory. This collaboration aims to accelerate everything from large-scale model training to industrial AI applications.
Decoded:
To power this expansion, Microsoft is deploying hundreds of thousands of NVIDIA Blackwell GPUs across massive new facilities in Wisconsin and Georgia that will serve its most demanding AI workloads.
The collaboration brings AI directly to enterprise data by integrating NVIDIA's open models and microservices with the new Microsoft SQL Server 2025, allowing businesses to run secure AI tasks where their data already lives.
Beyond data, the partnership is pushing into physical AI, enabling developers to use NVIDIA's Omniverse platform on Azure for transforming industrial workflows like digital twins and robotics simulation.
Why It Matters: This partnership is creating a deeply integrated ecosystem where AI hardware and software work together to lower costs and boost performance. By extending this optimized stack from cloud services to enterprise databases and physical robotics, the two companies are positioning AI as a core utility for the modern economy.
AI Pulse
Alphabet warned that the current AI investment boom has "elements of irrationality," with CEO Sundar Pichai stating that no company, including Google, would be immune if the bubble were to burst.
Cambridge Dictionary named 'parasocial' its 2025 Word of the Year, highlighting the rise of one-sided connections people form with celebrities, fictional characters, and increasingly, AI chatbots.
Anthropic launched its Economic Futures Program, a new initiative offering grants and policy forums to support research into AI's impact on labor markets and the global economy.
The International Labour Organization found in a new study that one in four jobs worldwide is potentially exposed to transformation by generative AI, with women in clerical roles facing a disproportionately high risk of disruption.