Blog
10.2025

Owning the Stack – Issue #3

This week, the AI race became a contest to own the entire stack. The largest technology companies are no longer focused on single breakthroughs. They are stitching together every layer of the AI value chain, from chips to data centers to consumer experiences.

Meta is finalizing a $30 billion financing for its Hyperion data center. OpenAI is co-developing custom chips with Broadcom. Google is investing $15 billion to build new infrastructure in India. Microsoft, NVIDIA, and BlackRock joined forces in a $40 billion acquisition of Aligned Data Centers. Apple launched the M5 chip to unify AI performance across its devices. The new frontier is not about who trains the largest model but who controls the full production line of intelligence.

The same shift is playing out inside the enterprise. Microsoft’s Copilot now operates across voice, vision, and actions. Salesforce introduced Agentforce 360 to let businesses build their own intelligent agents. Anthropic added new capabilities to Claude that bring structure and reliability to automation. Even Walmart connected ChatGPT directly to its shopping experience, blurring the line between enterprise and consumer AI.

Capital and research are following this vertical path. Fivetran merged with dbt Labs to connect data pipelines end to end. Revolut acquired an AI travel startup to own its customer journey from transaction to experience. DeepMind uncovered a breakthrough in cancer immunotherapy, showing how frontier science continues to translate into real-world impact.

Governments are beginning to respond in kind. California passed the first chatbot safety law. Europe committed one billion euros to its Apply AI initiative. The UK released new guidance to prepare its workforce for an AI-driven economy. The focus everywhere is shifting from experimentation to integration.

Here’s your Saturday guide to the signals shaping the future of AI:

Infrastructure

  • BlackRock, NVIDIA, and Microsoft finalize a $40 billion AI infrastructure consortium. The group, joined by Abu Dhabi’s MGX, will acquire Aligned Data Centers in one of the largest infrastructure transactions ever, setting a new benchmark for global AI capital formation. Click here
  • Meta nears a $30 billion financing for its Hyperion AI data center in Louisiana. The package would be the largest private capital commitment ever for a single data-center site, positioning Meta as one of the biggest infrastructure builders in the AI era. Click here
  • NVIDIA launches DGX Spark mini AI PCs for developers and donates Vera Rubin rack technology to the Open Compute Project. The move supports gigawatt-scale “AI factories,” expanding open access to high-efficiency computing architecture. Click here
  • OpenAI and Broadcom sign a multi-billion-dollar, four-year deal to co-develop 10 gigawatts of custom AI chips. TSMC will manufacture the chips, giving OpenAI a new compute supply chain independent of NVIDIA and AMD. Click here
  • Google commits $15 billion through 2030 to build a 1-gigawatt AI and data center hub in Andhra Pradesh, India. The project will be Google’s largest infrastructure investment in Asia and will help expand AI compute capacity across emerging markets. Click here
  • Intel previews its new “Crescent Island” AI data-center chip. Designed for high efficiency and flexible scaling, the chip enters production in 2026 and targets deployments that balance power use and performance for enterprise workloads. Click here

Enterprise

  • Microsoft rolls out major Windows 11 Copilot upgrades. The update adds voice activation, visual recognition, and new Copilot Actions, bringing more automation into everyday enterprise workflows. Click here
  • Salesforce debuts Agentforce 360, its expanded enterprise AI platform. The release enables companies to build, train, and deploy AI agents for customer service, operations, and data analysis with integrated security and compliance tools. Click here
  • Anthropic introduces Skills for Claude, allowing agents to perform pre-defined instructions and workflows. The new feature improves predictability and reliability for enterprise users working in regulated and complex industries. Click here
  • Walmart partners with OpenAI to integrate conversational shopping directly into ChatGPT. Shoppers can now browse, compare, and buy products in real time, marking one of the first mass retail applications of generative AI. Click here
  • Oracle launches AI Database 26ai, embedding intelligence directly into the database layer. It includes vector search, agentic automation, and quantum-resistant encryption, making AI-native data systems more secure and efficient. Click here
  • Apple unveils its M5 chip, a major leap in AI-optimized silicon. The new processor delivers four times the GPU compute performance of the M4 and strengthens Apple’s unified architecture for AI-driven devices. Click here

Capital Flows

  • Basis Theory raises $33 million to scale agentic AI infrastructure for payment processing. The round highlights growing investor appetite for secure, AI-driven financial systems. Click here
  • Dataiku surpasses $350 million in annual recurring revenue. The milestone underscores strong enterprise demand for AI platforms focused on governance, transparency, and risk management. Click here
  • Stand Insurance secures $35 million in Series B funding for AI home coverage. The company uses generative AI to underwrite policies in high-risk markets, signaling continued investor confidence in insurtech AI. Click here
  • Revolut acquires AI travel startup Swifty to expand its lifestyle automation ecosystem. The deal integrates conversational travel booking into Revolut’s 65 million user base, extending AI beyond finance into daily experiences. Click here
  • Fivetran and dbt Labs merge in an all-stock deal to form a $600 million ARR data powerhouse. The merger creates a unified pipeline from ingestion to transformation, enabling enterprise data systems to become fully AI-ready. Click here

Research

  • DeepMind’s C2S-Scale AI discovers a new pathway for cancer immunotherapy. The model identified a drug combination that boosts antigen presentation by fifty percent, which was later confirmed in human cell studies, demonstrating the power of large-scale predictive modeling. Click here
  • Microsoft unveils MAI-Image-1, its first in-house text-to-image model. The system ranks among the world’s best for lighting realism and visual accuracy, marking Microsoft’s deeper move into foundational model research. Click here
  • Anthropic publishes research on AI’s economic and policy implications. The study models multiple adoption scenarios and policy interventions to help governments manage labor and productivity shifts caused by automation. Click here

Policy

  • California signs SB 243, the nation’s first law regulating AI chatbot companions. The legislation requires transparency, user consent, and safety standards for emotionally interactive AI systems. Click here
  • Meta introduces new parental controls for AI characters on Instagram. Parents can now block AI interactions or monitor conversations, expanding the company’s approach to digital safety. Click here
  • Governor Gavin Newsom vetoes the “No Robo Bosses Act” in California. The governor cited overly broad restrictions on AI use in hiring and management, calling for more balanced regulation that supports both innovation and worker rights. Click here

Global AI Strategy

  • China now leads global open-source AI model development. All top-performing open models currently originate in China, reversing last year’s US dominance and signaling a strategic shift in global AI leadership. Click here
  • ByteDance’s Doubao app becomes China’s leading consumer AI platform. The success reflects rapid adoption, national governance oversight, and deeper integration with Douyin’s social ecosystem. Click here
  • The UK releases “The Engineer of the Future,” a new national AI workforce plan. The framework outlines how AI will redefine engineering roles and digital capabilities in the public sector. Click here
  • Microsoft and OpenAI launch international AI training programs for public servants. The initiative aims to standardize ethical deployment and governance education across more than 100 countries. Click here

To go deeper, subscribe to my monthly Founder Insights newsletter, where I share lessons from the frontlines of company building, perspectives on AI’s future, and our industry’s road ahead.

Originally published on LinkedIn.

# #