The White House launches an AI effort to support innovation, Gemini 3 powers Google upgrades, and Anthropic replies to last week’s launches and commits to U.S. AI infrastructure.

Plus, Sam Altman and an Apple design legend give a glimpse into the future of AI devices.

Here’s the rundown:

1. White House launches ‘Genesis Mission’ to accelerate innovation with AI

President Trump signed an executive order creating the Genesis Mission, a federal initiative to accelerate scientific discoveries in materials engineering, health sciences, and energy. The plan seeks to use the computing resources of the Department of Energy’s national labs to tap federal datasets for more AI-driven experiments. Private-sector partners, such as NVIDIA, AMD, Dell, and HPE, will boost computing resources at the labs. The initiative also aims to lower production and energy costs by improving grid efficiency in response to growing concerns over the power demands of present AI infrastructure.

CFO takeaway: A few weeks ago, we discussed how Google and Yale used AI to generate a novel cancer therapy hypothesis, shifting the bottleneck from discovery to validation. The Genesis Mission helps address that gap by enabling earlier simulation, modeling, and testing with access to federal compute and data. Validated discoveries can move faster to commercialization, so companies will need to front-load capital allocation to keep up with validated breakthroughs.

Read more

OpenAI

2. OpenAI hints at ‘simple,’ everyday AI device

At Emerson Collective’s Demo Day, famed Apple designer Jony Ive revealed he has built a working prototype of an AI device in collaboration with OpenAI. The project is currently in stealth, with a public launch expected in less than two years, Ive said. Altman described the device’s design as playful, simple, and beautiful, while Ive emphasized his love of creating intuitive products that feel unintimidating and effortless to use. The project follows OpenAI’s $6.5B deal with Ive’s AI hardware startup, whose team now forms OpenAI’s internal hardware division.

CFO takeaway: AI builders are already racing against time to keep up with development, and the features OpenAI ships set a benchmark. By potentially launching a new product category, Altman and Ive have cranked up the pressure in an already unprecedented AI evolution cycle. It’s another sign that Altman wants to dream big and help set the future of consumer technology.

Read more

Anthropic

3. Anthropic launches Claude Opus 4.5

Anthropic released Claude Opus 4.5, with improved performance on real-world coding, agent, and productivity tasks. Anthropic says Opus 4.5 outperformed all human candidates on its internal coding test and demonstrated creative reasoning on complex, multi-step benchmarks. The update powers new tools including Claude for Excel, Claude for Chrome, and a desktop version of Claude Code with multi-agent coordination, longer memory, and smarter context handling. Pricing remains $5 per million input tokens and $25 per million output tokens, and the model is now live across all major cloud platforms.

CFO takeaway: LinkedIn News called Claude 4.5 Anthropic’s answer to the latest releases from OpenAI and Google. It’s another sign the competition around who will power engineering and agentic workflows is heating up. To compete, Anthropic is positioning its models to cover every stage of that pipeline: Opus for lead agents and production code, Sonnet for iteration, and Haiku for sub-agent tasks. As CPO Mike Krieger noted, “The Claude 4.5 generation spans the full development lifecycle.”

Read more

Google

4. Google upgrades Nano Banana with Pro model

Google has released Nano Banana Pro, an upgraded version of its image model with sharper text rendering, the ability to pull in the latest information from Search, and studio-level editing tools like lighting control, depth-of-field, and multilingual generation. It can generate complex scenes, maintain character consistency, and localize visuals for global use. The original Nano Banana remains available for what Google calls casual editing, while the Pro version is aimed at more advanced creative work. It’s rolling out across Google Ads, Slides, Search, NotebookLM, and Vertex AI.

CFO takeaway: After releasing Gemini 3 last week, Google is showing how the model powers upgrades across its suite of products. And as we noted in the Anthropic story above, the battle among AI giants is focused on owning the workflows of technical teams. Nano Banana’s upgrade is part of that mission. Sundar Pichai summed it up on X: “Built on Gemini 3, it’s really good at complex infographics — much like how engineers see the world.”

Read more

Microsoft

5. Microsoft Copilot rolls out AI agents across Office and Teams

Microsoft introduced AI agents across Word, Excel, PowerPoint, Outlook, and Teams at its annual Ignite enterprise tech conference. The agents are powered by Work IQ, an intelligence layer developed by Microsoft that gives agents company context by learning from emails, meetings, files, and workflows. A new control hub, Agent 365, lets organizations manage access to data, monitor usage, and enforce security. Copilot can now run on models from OpenAI and Anthropic, and a new partnership with Harvard’s D^3 Institute will help executives adopt agent-led workflows.

CFO takeaway: Like OpenAI’s Company Knowledge, Microsoft is adding organizational context to optimize AI’s performance. But it’s going a step further with agents that can act on users’ behalf within everyday workflows. That shift makes it even more important for companies to have structured data sources and companywide sources of truth so agents can act on accurate, up-to-date information.

Read more

Amazon

6. Amazon commits up to $50B to expand AI infrastructure for U.S. government

Amazon plans to invest up to $50 billion to expand AI and supercomputing infrastructure for U.S. government agencies. The buildout, set to begin in 2026, will add 1.3 gigawatts of compute capacity across AWS’s GovCloud, Secret, and Top Secret regions — secure cloud environments designed for handling classified and sensitive government data. Agencies will get access to tools like SageMaker (for training models), Bedrock (for deploying models and agents), Trainium chips, NVIDIA GPUs, and foundation models like Anthropic Claude. This will facilitate faster simulation, threat detection, and decision-making across national security, science, and infrastructure.

CFO takeaway: Amazon has long supported secure cloud for the U.S. government. With this investment, it’s moving beyond hosting data to powering the AI infrastructure that will drive critical government missions. And by embedding AI and supercomputing directly into classified environments, government agencies can run autonomous discovery and threat detection at unprecedented speed. For AWS, it’s a defensible advantage: few players have both the model partnerships and the deeply accredited infrastructure across Secret and Top Secret regions. For the government, it opens up potential productivity breakthroughs.

Read more

See you next week, and happy Thanksgiving!

In other news

1. AI music startup Suno raises $250M at a $2.45B valuation (Suno)
2. OpenAI to sunset GPT-4o API by February 2026 (OpenAI)

2 new AI tools to try

1. GPT-5.1-Codex-Max (OpenAI)
2. NotebookLM: Slide Decks and Infographics (Google)

Recommended reading

1. Altman memo forecasts ‘rough vibes’ due to resurgent Google (The Information)
2. Early experiments in accelerating science with GPT-5 (OpenAI)