Trump Signs Sweeping AI Executive Order

President Trump signed a sweeping Executive Order today that fundamentally reshapes the US government's approach to AI. For the first time, companies developing advanced AI models are asked—though not yet mandated—to voluntarily share those models with the government for up to 30 days before public release. This marks a major shift from the previous laissez-faire stance that characterized the early AI boom. The order also establishes an AI cybersecurity clearinghouse, a centralized body to coordinate threat intelligence and response across federal agencies and private sector partners. The Attorney General is directed to prioritize enforcement actions against criminal uses of AI, including deepfake fraud, automated cyberattacks, and AI-generated disinformation. Industry reaction is mixed: some see it as necessary guardrails, others as a chilling effect on innovation. The voluntary nature leaves room for interpretation, but the signal is clear—the government wants a seat at the table before your model goes live.

Microsoft Build 2026 delivered a bombshell: Project Polaris, Microsoft's own AI coding model, will replace GPT-4 in GitHub Copilot by August 2026. This is a strategic decoupling from OpenAI, signaling that Microsoft believes its in-house models are now competitive. The move comes alongside the launch of Windows Agent Framework 1.0, a platform for building and running AI agents directly on Windows. Azure Agent Mesh enables federated multi-agent execution, allowing agents to collaborate across cloud and edge. Copilot Workspace reached General Availability, and Agent Mode became the default in Office 365 Copilot, meaning every Office user now has an AI agent baked in. The crown jewel is MAI-Code-1-Flash, Microsoft's first code generation model, optimized for low-latency, high-throughput coding tasks. This is the most aggressive AI push from Microsoft yet, and it directly challenges OpenAI, Google, and Anthropic on their own turf.

The Department of Defense announced a $9.69 billion contract with Microsoft, the largest single government contract in the company's history. The deal covers a comprehensive suite of Microsoft 365, Azure cloud services, and AI-powered Copilot tools across the DoD's vast infrastructure. The DoD projects $422 million in savings through efficiency gains, automation, and reduced licensing costs. This contract solidifies Microsoft's dominance in government cloud and AI, especially after the controversial JEDI contract cancellation years ago. For the DoD, it's a bet that AI-powered productivity tools will modernize operations, from logistics to cybersecurity. The contract includes strict compliance with FedRAMP and IL5 security standards, ensuring classified workloads can leverage AI. Expect other federal agencies to follow suit, making Microsoft the de facto AI provider for the US government.

Alphabet announced a massive $80 billion capital raise dedicated entirely to AI infrastructure. The most eye-catching detail: Berkshire Hathaway, Warren Buffett's investment vehicle, is writing a $10 billion check. This is a seismic signal from the Oracle of Omaha, who has historically avoided tech bets. The funds will go toward expanding Google's data center fleet, building new AI-optimized chips (TPU v6), and securing long-term energy contracts for compute. Alphabet is essentially building a national-scale AI computing grid, rivaling anything from Microsoft or Amazon. The move comes as AI model training costs continue to skyrocket, with frontier models now costing billions. Buffett's involvement suggests he sees AI compute as a long-term, defensible asset class—like railroads or utilities. For startups, this means the compute gap will widen; you'll either need to partner with hyperscalers or find specialized niches.

NVIDIA and Microsoft unveiled the RTX Spark superchip at Microsoft Build 2026, marking a paradigm shift in personal computing. The chip delivers 1 petaflop of AI performance, equivalent to a small data center, with up to 128GB of unified memory. This enables running large language models, diffusion models, and AI agents entirely on-device, without cloud dependencies. The RTX Spark is designed specifically for personal AI agents—autonomous assistants that live on your PC, manage your files, schedule, and communications. Microsoft's Windows Agent Framework 1.0 is optimized to leverage this hardware, creating a seamless local AI experience. For enterprises, this means sensitive data never leaves the device, solving compliance and privacy headaches. For consumers, it means AI that works offline, instantly, and without subscription fees. The first RTX Spark PCs are expected by September 2026, starting at $2,999.

Anthropic confidentially filed its S-1 with the SEC on June 1, setting the stage for what could be the largest tech IPO in history. The company is valued at $965 billion following a $65 billion Series H round, with a revenue run rate of $47 billion—a staggering number that surpasses OpenAI's reported $37 billion. Anthropic's Claude models have become the default for enterprise AI, especially in regulated industries like healthcare, finance, and law. The IPO filing reveals that Anthropic has secured long-term compute agreements with Google and AWS, ensuring capacity through 2030. The company is also investing heavily in AI safety research, a differentiator that resonates with risk-averse enterprise buyers. Getting ahead of OpenAI, which is also pursuing an offering, gives Anthropic first-mover advantage in the public markets. The roadshow will likely emphasize Anthropic's enterprise focus and safety-first approach as key differentiators.