The Public Reckoning

Anthropic filed for IPO yesterday at a $965 billion valuation -- more valuable than OpenAI for the first time. OpenAI is weeks behind with its own filing. The two largest AI infrastructure companies are about to face public market scrutiny simultaneously. For enterprise leaders who signed SaaS agreements with AI startups, the vendor relationship is changing permanently.

In this edition: Anthropic's $965B IPO filing  ·  Cognition's Devin: 13x revenue growth  ·  Goldman Sachs and U.S. military running AI agents in production  ·  Novo Nordisk goes all-in on OpenAI  ·  Google Gemini Agent Platform  ·  Cisco Live keynote today

Your AI Vendor Is Going Public

Anthropic filed a confidential S-1 with the SEC on June 1 -- the day after raising $65 billion at a $965 billion valuation. OpenAI is preparing its own filing. Both are approaching $1 trillion on the back of enterprise revenue. The commercial relationship between enterprise buyers and AI vendors is about to undergo the most significant structural change in its short history.

The headline is historic: an AI safety company founded four years ago is now more valuable than its chief rival, heading to public markets ahead of it. But the enterprise-facing implication is distinct from the venture capital story, and it is the one that matters most to the leaders reading this.

When Anthropic goes public, enterprise customers gain something they currently lack: financial transparency. An S-1 will document revenue concentration, customer retention rates, contract structures, and the commercial dependencies that currently exist only in signed agreements. Public company status introduces earnings calls, analyst scrutiny, and quarterly pressure that disciplines how large vendors manage customer relationships, pricing, and product roadmaps. It also introduces risk: a $965 billion company carries public market expectations that a privately-funded safety research organization does not. Enterprise buyers who currently negotiate with a company that is mission-driven first and revenue-driven second are negotiating with a different kind of company once those priorities are publicly accountable to shareholders.

OpenAI is right behind. The company raised $122 billion at an $852 billion valuation and is preparing its own IPO filing. Enterprise revenue now accounts for more than 40% of OpenAI's total, on track for parity with consumer by year-end. Two trillion-dollar AI companies going public in the same window means the enterprise AI vendor market will face public scrutiny simultaneously -- revenue concentration, customer churn, pricing power, and competitive positioning will all become readable in the same reporting cycle.

For enterprise CIOs and CPOs, the S-1 filing is a procurement signal. The terms your legal team negotiated with a $50 billion private startup carry different implications when the counterparty is a public company with institutional shareholders and Wall Street analysts asking quarterly about revenue per enterprise customer. Long-term contracts gain new value as pricing predictability becomes a competitive moat. Strategic consolidation on fewer platforms reduces the complexity of managing vendor relationships at public-company scale.

Moving Pieces

Four developments that matter to enterprise leaders this week

Cognition's Devin: 13x Revenue Growth, Goldman and the U.S. Military Running It in Production

Cognition AI raised more than $1 billion in a Series D round (co-led by Lux Capital, General Catalyst, and 8VC) at a $26 billion valuation. Revenue run rate grew from $37 million in May 2025 to $492 million by May 2026 -- a 13x increase in 12 months. Enterprise customers running Devin, the autonomous AI coding agent, in production include Citi, Goldman Sachs, Mercedes-Benz, Elevance, Dell, Santander, the U.S. Army, and the U.S. Navy. Enterprise usage has grown more than 10x since January 2026 alone. Devin now ships 89% of Cognition's own production code. The Cognition story is the clearest real-world data point yet on what agentic AI coding looks like at scale inside regulated, high-stakes enterprise environments -- and it is growing fast enough that it will be a public company conversation within two to three years at this trajectory.

Novo Nordisk Deploys OpenAI Across Its Entire Operations

Novo Nordisk announced a strategic partnership with OpenAI to integrate AI across its entire business -- from drug discovery and clinical trials to manufacturing, supply chains, and commercial operations -- with full deployment planned by end of 2026. This is not an AI pilot in one business unit. It is an enterprise-wide commitment by a $500 billion pharmaceutical company to rebuild how it finds, develops, and commercializes drugs. OpenAI will also build AI fluency across Novo Nordisk's global workforce. For enterprise leaders in regulated industries watching AI adoption in pharma: this is the most significant full-enterprise production commitment from a major pharmaceutical company to date, and it is structured with strict data protection, governance, and human oversight requirements baked in -- not as a compliance add-on, but as a precondition of the partnership.

Google's Gemini Agent Platform: Managed Agents and Governance Built In

Google's Gemini Enterprise Agent Platform, unveiled at Google Cloud Next and expanded at Google I/O, allows developers to build and run custom AI agents inside secure, Google-hosted environments where agents automatically inherit enterprise-grade data privacy, governance, and security. Gemini Spark is a 24/7 personal AI agent for Gemini Enterprise and Workspace customers that autonomously acts on users' behalf. CodeMender is an AI security agent that finds and fixes code vulnerabilities through the Agent Platform. The enterprise governance thesis is pointed: governance friction is the primary adoption barrier for agentic AI, and removing it at the infrastructure layer -- rather than requiring per-deployment configuration -- changes the deployment calculus. Google is betting that managed, compliance-first agent infrastructure is the architecture most enterprise security teams will accept, even if it trades some flexibility for control.

Cisco Live Day 2: Chuck Robbins on the Agentic Network

Cisco CEO Chuck Robbins headlines the Cisco Live Las Vegas keynote today with the company's $9 billion AI infrastructure order target and its AgenticOps framework as the central narrative -- networks that detect, diagnose, and remediate issues without human intervention at every step. The Astrix integration extends Zero Trust to non-human identities: the API keys, service accounts, and OAuth tokens operated by AI agents inside enterprise systems. Cisco's framing of the enterprise AI challenge as three simultaneous constraints -- infrastructure capacity, trust and governance, and data accessibility -- is a useful map for enterprise leaders prioritizing investment. Most organizations have been investing heavily in the first. The second and third are where most deployments stall. AgenticOps addresses the first; the security framework addresses the second. The data layer remains the hardest problem in the stack.

Anthropic's annual revenue run rate at the time of its IPO filing -- up from approximately $3 billion 18 months ago.

When Anthropic filed a confidential S-1 with the SEC on June 1, its run-rate revenue had crossed $47 billion -- driven overwhelmingly by enterprises embedding Claude in production operations. That growth trajectory is what supports a $965 billion valuation and a near-trillion-dollar IPO. The number matters for enterprise leaders not as a vendor milestone but as a structural data point: the AI vendor market has grown large enough, fast enough, that the companies at its center are entering public markets. The commercial terms of enterprise AI are about to be set in quarterly earnings calls rather than venture-backed sales negotiations. For procurement teams, that is a different environment entirely.

Sources: Anthropic Series H  ·  Fortune  ·  May 28 -- June 1, 2026

The Vendor Relationship You Didn't Know You Had

There is a version of enterprise AI procurement that treats AI vendors the way software teams have always treated SaaS vendors: annual renewal, price cap, SLA, exit clause. That version is no longer adequate.

When Anthropic goes public, the S-1 it files will contain more information about the commercial structure of AI vendor relationships than enterprise procurement teams have ever been able to access. Revenue concentration by customer, gross margins, contract duration, churn rates -- all of it becomes readable. Enterprise leaders who have been making AI vendor decisions on the basis of model benchmarks and sales conversations will be able to read the actual unit economics for the first time.

The Cognition numbers tell the other side of this story. Thirteen times revenue growth in 12 months. Goldman Sachs, Citi, the U.S. Army running an autonomous coding agent in production. These are not pilot customers. They are enterprise organizations that have made material bets on AI infrastructure they do not control, provided by a company now worth $26 billion. The dependency is real. The procurement frameworks most enterprises built to manage it are not sized for this.

The practical work is not complicated, but it requires time most procurement teams have not reserved for it. Map your AI vendor dependencies as you would any critical infrastructure supplier: what is the switching cost, what happens to production operations if terms change, and which of your AI deployments are production systems rather than pilots? Production systems require vendor relationships governed at the infrastructure tier. The IPO filings will tell you whether the vendor you chose can sustain the relationship. Read them.

AI vendors are no longer startups negotiating flexibility. They are public-market companies negotiating from permanence. Adjust your procurement posture accordingly.

AK  /  Spearhead  /  Building AI systems, not tools