Blog
11.2026

The Enterprise AI Playbook – Issue #53

Spotlight: The Enterprise AI Playbook

Enterprise AI is moving from use cases to business transformation.

That was the clearest signal from @JPMorgan Chase’s annual Tech Summit, hosted by @Larry Feinsmith, Head of Technology Strategy, Innovation & Partnerships. Global Chief Information Officer @Lori Beer described the shift from counting roughly 1,000 use cases to tracking 50 strategic AI initiatives, prioritizing business transformation over activity volume. Their ambition is to move from idea to production in four months. In one of the most regulated industries, that pace sets a high bar.

For founders, here’s what it takes to be enterprise-ready:

1. Sell outcomes, not tokens. Economics is moving from token budgets to cost per task, including the costs of building and running.

2. Go headless. Products need APIs and MCP interfaces to take part in agent-driven workflows. As orchestration moves up to the agent, the interface becomes intent.

3. Fit the customer’s AI factory. Your product has to work inside the architecture and controls the enterprise has chosen.

4. Treat security as the control plane. Unique agent identities, just-in-time access, and runtime guardrails are becoming requirements. Governing agent-to-agent interactions is still unsolved, and that may be a big opportunity.

Enterprise complexity is routinely underestimated. The startups that show fully loaded economics and deploy cleanly inside that complexity will earn the right to scale.

Full Weekend Edition below. 👇

Signals Shaping the Future of AI:

Infrastructure

  • SpaceX announced plans to invest more than $74 billion in AI supercomputing campuses in Texas and Arizona. The proposed projects include a roughly $24 billion Starbase facility with on-site power generation and a $50 billion Arizona development that still requires approvals. Click here
  • Anthropic agreed to purchase approximately $3.4 billion of Caterpillar power-generation equipment for its AI data centers. The 1.7-gigawatt order underscores how frontier AI developers are securing dedicated electricity supplies to support compute expansion. Click here
  • Google introduced TPU 8tx and TPU 8i Max, expanding its eighth-generation AI chip portfolio. The systems target confidential inference and larger-context agentic workloads, respectively, as Google broadens its alternatives to NVIDIA hardware. Click here
  • CoreWeave said it expects approximately $9 billion in customer prepayments and deposits over the next three months to fund AI infrastructure expansion. Third-quarter revenue reached $3.66 billion, up 168%, while its contracted backlog climbed to $132.5 billion. Click here
  • SoftBank canceled 13 gigawatts of turbine reservations for AI data centers, taking a $3.7 billion earnings hit. The company had reserved more power-generation equipment than its planned projects required, highlighting the risks of aggressive infrastructure commitments amid the AI buildout. Click here

Enterprise

  • Anthropic released Claude Fable 5.2 and Claude Mythos 5.2, improving advanced reasoning and coding performance while cutting prices by 20%. The models generate outputs up to 30% faster, with Mythos access restricted to vetted organizations working in sensitive fields. Click here
  • Databricks upgraded its Genie AI agent to autonomously execute longer, multistep business tasks. The always-on agent uses enterprise data and organizational context to handle analysis and decision-support workflows with usage controls. Click here
  • United Airlines moved to block unauthorized third-party AI agents from booking flights and managing reservations. The carrier cited cybersecurity, customer data exposure, and regulatory risks as businesses confront the question of how autonomous agents should access their systems. Click here
  • Fireworks AI confirmed a security breach that exposed customer data, including some Notion AI chat transcripts. The incident affected fewer than 1% of Notion workspaces and highlighted the risks enterprises face when relying on third-party AI infrastructure. Click here
  • Google released Gemini 4 Hydrogen, a faster model optimized for everyday coding and agentic workflows. The model extends the Gemini 4 family to lower-cost, high-volume applications and offers multiple reasoning settings for performance and efficiency. Click here

Capital Flows

  • Anthropic set its IPO price range at $70 to $78 per share, implying a fully diluted valuation of up to $1.404 trillion. The offering could raise up to $64.35 billion, alongside a separate $26 billion private placement involving major technology suppliers. Click here
  • OpenAI reportedly raised $15 billion at a $1.25 trillion valuation, about half the amount it initially sought. The CNBC-reported financing comes as investors weigh the enormous capital needs and valuations of frontier AI developers. Click here
  • Mirendil raised $1 billion in a Series A to build AI systems that accelerate frontier research and development. The startup is developing tools that allow teams of AI agents to autonomously write code, conduct experiments, and train models. Click here
  • Mercor raised $750 million at a $20 billion valuation to expand AI training and evaluation. The company connects AI developers with domain experts to train and test agents on complex professional tasks. Click here
  • LangChain raised $300 million at a $3 billion valuation, more than doubling its previous valuation. The financing will support its platform for building, monitoring, and deploying enterprise AI agents across different models and systems. Click here

Research

  • Google DeepMind introduced AlphaProof Code, an AI framework that mathematically verified key security properties of the Signal encryption protocol. The system generated machine-checked proofs of important cryptographic guarantees, demonstrating how AI can help verify complex real-world software. Click here
  • Microsoft researchers demonstrated that editing fewer than 10 neurons in an AI model can create hidden backdoors in autonomous agents without retraining. The technique, called Lazarus, can cause compromised agents to execute malicious commands when triggered while still passing standard safety evaluations. Click here
  • Nearly 6 in 10 enterprise executives plan to increase their use of open-weight AI models to manage costs, according to Capgemini research. The survey found that 59% had exceeded AI spending expectations, reinforcing cost efficiency as a central consideration in model selection. Click here
  • Google DeepMind unveiled AlphaGenome 2, a model that predicts how genetic variants affect specific human cell types. Trained on more than 100 million cell profiles across 68 tissues, it identifies nearly twice as many gene-expression-changing variants as its predecessor. Click here

Policy

  • The U.S. Senate voted 73-25 to advance legislation that would accelerate permitting for energy infrastructure and data centers. The bipartisan proposal seeks faster project approvals while requiring large data center operators to cover their share of electricity infrastructure costs. Click here
  • The U.S. Commerce Department is reportedly finalizing rules that would restrict Chinese AI firms from remotely accessing advanced chips via overseas data centers. The proposal could expand semiconductor controls to seven additional countries, including Malaysia, Singapore and Thailand. Click here
  • New York opened mandatory registration for major frontier AI developers under its RAISE Act. Covered companies must register by January 1, 2027, and comply with requirements for public safety frameworks, transparency reports and critical-incident reporting. Click here
  • Federal Reserve Governor Michael Barr warned that rapid AI adoption could cause significant near-term labor disruption. He cautioned that insufficient guardrails could worsen inequality as businesses adopt increasingly capable AI systems. Click here
  • Romulus, Michigan, imposed a 180-day moratorium on new gigawatt-scale data center proposals. The decision illustrates growing local resistance to large AI infrastructure projects as communities weigh electricity demand and land-use impacts. Click here

Global AI Strategy

  • UK-backed AI chip startup Fractile raised $720 million at a valuation of approximately $7 billion. Britain’s sovereign AI investment program participated in the round, supporting domestic inference hardware as the country seeks greater control over critical AI infrastructure. Click here
  • Seventeen governments and ten AI developers endorsed the Paris Call for AI Safety and Security on November 9. Signatories including OpenAI, Anthropic and Mistral backed international cooperation on fail-safe mechanisms, independent testing and early warnings about serious AI incidents. Click here
  • NVIDIA, Rebellions and SK Telecom announced plans to integrate South Korean AI chips with NVIDIA’s NVLink Fusion infrastructure. The collaboration aims to support domestic AI computing capacity and sovereign services using Korean-designed processors and models. Click here
  • Saudi Arabia’s National Infrastructure Fund agreed to provide HUMAIN with up to $667 million in financing for AI infrastructure. The agreements will support domestic GPU deployment and power facilities as the kingdom expands its sovereign computing capacity. Click here
  • ByteDance is reportedly preparing a personal AI agent to compete with Meta’s Muse and leading U.S. AI assistants. The planned launch reflects China’s push to build consumer-facing AI products alongside domestic model development. Click here

Talent Signals

Each week, we spotlight key roles tied to the themes shaping this week’s AI headlines, connecting talent to the companies driving the news.

  • DevRev is building an AI-native platform that connects enterprise knowledge, customer support, and business operations through intelligent agents. Following the expansion of its Computer and Operator offerings, the Mayfield-backed company is hiring across applied AI engineering, forward-deployed solutions, product, design, and enterprise go-to-market roles. Click here
  • Rox is building AI agents that run revenue workflows for enterprise sales teams. Following its $100 million Series B at a $1.3 billion valuation, Rox is hiring across engineering, go-to-market, growth and marketing roles. Click here
  • UFORCE is building AI-enabled autonomous defense systems across air, land, and sea, combining deployable hardware with software for mission planning and coordination. The company is hiring across AI engineering, autonomous software, platform development, operations, and business development. Click here

You can see all the opportunities at Mayfield-backed AI companies here.

Social Signals

The most important conversations in AI are unfolding across social media, where top voices are shaping the next wave of signals and strategy. Here are some of the top social signals and their takes from the past week.

  • Matt Huang (Click here): “There seems to be a new class of people, call them ‘swarm riders,’ who are capable of consistently commanding large swarms of agents to productive ends. This is a new kind of managerial skill, analogous to the skill it takes to manage large organizations of people, but wholly different in nature and of course more technical.” Huang sees swarm orchestration becoming a distinct operating skill, one built around reward functions, evals, experimentation, and designing work that agents can execute with minimal friction. If that advantage proves durable, “swarm-native” companies may look less like traditional organizations with AI layered on top and more like a genuinely new organizational form.
  • Ethan Mollick (Click here): “I had early access to Fable 5.2 for a bit & I think it crosses the line into being a sort of Jagged AGI. Here, I have it launch a business, do academic research, hack hardware, build 3D models & more. I think it is a good illustration of limits & powers.” Mollick uses “Jagged AGI” to describe a system that can perform impressively across very different kinds of work while still failing unpredictably in others. The practical takeaway is less about whether AI has crossed a formal AGI threshold and more about how quickly the range of tasks worth testing with frontier models is expanding.
  • Sundar Pichai (Click here): “Our teams put Gemini 4 Carbon to work on a major security challenge: mathematically proving the code underlying the Signal protocol. Gemini 4 Carbon generated over 6M lines of machine-checked Lean proofs, compressing years of expert manual verification into just weeks.” This is a striking example of AI being applied not just to write software, but to prove that software behaves as intended. If the approach scales, formal verification could move from specialized, labor-intensive work toward a much more practical security layer for critical software systems.

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: https://www.linkedin.com/newsletters/founder-insights-7274531066957217793/

↓ Drop a note in the comments with the areas of AI you want us to explore next.

Originally published on LinkedIn.

# #