At its annual conference in Las Vegas, ServiceNow announced an expanded AI Control Tower, 20 new Autonomous Workforce AI Specialists, Action Fabric for governed agent interoperability, and Autonomous Security & Risk. Then it offered AI Control Tower free for one year to any enterprise. That last move is the one that matters most.

The week began with OpenAI and Anthropic each announcing private equity joint ventures to accelerate enterprise distribution. It ends with ServiceNow announcing that if you want AI governance — the layer that sits above all those distributed agents — you can have it at no cost for the next twelve months. The sequencing is not coincidental. The governance layer is where the enterprise AI market is now being contested, and ServiceNow is willing to give it away to win the relationship.

The expanded AI Control Tower can now discover, govern, observe, secure, and measure every AI agent, model, and workflow deployed across an enterprise — regardless of who built it or which cloud it runs on. It integrates with AWS, Microsoft Azure, Google Cloud, NVIDIA, Anthropic, OpenAI, and more than thirty other connected systems.

The Autonomous Workforce expansion is the product that will generate the most near-term procurement conversations. ServiceNow has launched 20 role-scoped AI Specialists across IT operations, CRM, HR, finance, legal, procurement, and security and risk. These are autonomous agents that complete end-to-end processes: the L1 IT Service Desk AI Specialist is resolving cases 99% faster than human agents in ServiceNow's own help desk. DocuSign is targeting autonomous resolution of 90% of all IT tickets. Honeywell's AI assistant has eliminated the majority of service desk conversations. The city of Raleigh reports a 98% deflection rate on employee requests, saving the equivalent of a full month of staff time.

The most architecturally significant announcement is Action Fabric. ServiceNow has accumulated twenty years of enterprise workflows, approval chains, and business rules running across 100 billion workflows annually. Action Fabric now exposes all of that to any AI agent — whether built on ServiceNow, Claude, Microsoft Copilot, or a customer's own stack — through REST APIs, MCP, or agent-to-agent protocols. Every action taken through Action Fabric inherits the same governance, audit logging, and compliance controls as if it were executed by a ServiceNow workflow. Anthropic is the first launch partner: Claude can now connect directly to ServiceNow's governed system of action.

The free AI Control Tower offer deserves its own analysis. ServiceNow is offering any enterprise that signs up today twelve months of AI Control Tower access at no cost — a stated $2 million value. The strategic logic is explicit: if ServiceNow owns the governance relationship, every subsequent procurement conversation about autonomous workflows, AI specialists, and security risk management starts inside the ServiceNow platform. The free year is the land. Action Fabric is the expand. The Autonomous Workforce is the lock-in.

ServiceNow also set a 2030 financial target: $30 billion or more in subscription revenues, with AI expected to represent more than 30% of annual contract value — implying over $9 billion in AI-specific ACV by 2030.

Sources: ServiceNow official Knowledge 2026 release, May 5, 2026 / Fortune, May 5, 2026 / Diginomica, May 5, 2026 / Shashi.co analysis, May 5, 2026

the stated value of the AI Control Tower offer ServiceNow is giving away free for one year to any enterprise committing today.

Enterprise software companies do not give away $2 million worth of product without a very specific reason. ServiceNow's reason is that the governance relationship — the platform where CIOs and CISOs go to see every AI agent, every model, every permission, every workflow — is worth more than $2 million in Year 2 retention value. The free year establishes the integration, the trust, and the data context that makes switching costs prohibitive. It is the classic enterprise SaaS land-and-expand, applied to the most strategically important layer in enterprise AI. ServiceNow has 25,000 attendees at Knowledge 2026 and 100 billion workflows running annually. It is betting those two facts compound into $30 billion in subscription revenue by 2030.

Action Fabric exposes ServiceNow's two decades of enterprise workflows, approval chains, business rules, and operational playbooks to any AI agent — Claude, Copilot, Gemini, or a customer's own stack — through REST APIs, MCP, or agent-to-agent protocols. Every action inherits the same governance, audit logging, and compliance controls as native ServiceNow workflows. Anthropic is the first launch partner: Claude connects directly to ServiceNow's governed system of action. For enterprises building multi-vendor AI stacks, Action Fabric is the interoperability layer that makes governance possible without requiring everything to live inside ServiceNow.

Sources: ServiceNow Newsroom, May 5 / Shashi.co analysis

ServiceNow's $7.75 billion Armis acquisition is showing up in production form at Knowledge 2026. Autonomous Security & Risk integrates Armis's continuous asset intelligence (tracking nearly 7 billion connected assets across IT, OT, IoT, and cloud workloads in real time) with Veza's access graph (mapping every human and non-human identity and permission). For the first time, security teams have a unified picture of what exists and who — or what AI agent — is permitted to interact with it. As enterprises deploy more AI agents, each becomes a non-human identity with permissions. Veza's access graph now maps AI agent identities alongside human identities. The security business crossed $1 billion in ACV last year.

Sources: CX Today, May 6, 2026 / Diginomica, May 5

ServiceNow and Microsoft announced today a deepened integration extending ServiceNow AI Control Tower governance across the full Microsoft Agent 365 ecosystem — beyond Foundry and Copilot Studio to the entire Agent 365 agent ecosystem. ServiceNow AI Specialists will be available in the Microsoft Agent 365 Marketplace, appearing in the org chart as digital employees with defined roles, permissions, and accountability. The integration is available in preview today. For enterprise IT teams running Microsoft 365 environments, ServiceNow's governance layer now extends across every AI agent operating in their Microsoft environment. The two governance platforms layer rather than compete: Agent 365 manages identity, AI Control Tower manages behavior, permissions, and audit trail.

Sources: ServiceNow Newsroom, May 6, 2026 / CXO Today, May 6

ServiceNow and NVIDIA announced Project Arc at Knowledge 2026 — an enterprise autonomous desktop agent secured by the NVIDIA OpenShell sandboxed runtime. Project Arc operates on the employee desktop, executing tasks across applications within a governed, policy-constrained, auditable environment. For enterprise security teams concerned about AI desktop agents creating unaudited access to sensitive local data and applications, Project Arc is the first production-grade sandboxed desktop agent from major enterprise software vendors.

Sources: Diginomica, May 5, 2026 / NVIDIA OpenShell framework announcement

ServiceNow's free AI Control Tower offer is the most audacious enterprise software pricing move since Salesforce established its market position through aggressive pricing in the mid-2000s. It is also a move that has a specific financial logic that enterprise buyers should examine carefully.

Enterprise software platforms do not become free when they run out of value. They become free when the vendor's long-term retention value exceeds the short-term revenue from charging for the product. The integration depth that makes AI Control Tower valuable — agent discovery, CMDB connections, Armis asset intelligence, Veza identity graph — is also the integration depth that makes switching prohibitively expensive after Year 1.

For enterprise buyers, the right question is not "how do we get this for free" but "what does our relationship with ServiceNow look like after the free year ends, and what is the cost of switching governance platforms once our AI estate is mapped to their CMDB?" Enterprise procurement teams that do not negotiate Year 2 pricing before accepting the Year 1 offer are making a structural mistake.

That said: the governance gap is real, the 64% of engineers who deployed agents before their organizations were ready is real, and if ServiceNow can genuinely close that gap faster and more reliably than building in-house — the free offer deserves serious evaluation. The caution is about negotiation discipline, not about the product's value.

Look at what happened in the past two weeks. Microsoft shipped Agent 365 with built-in governance primitives. Google launched Gemini Enterprise Agent Platform with Agent Gateway, Agent Registry, and Agent Identity. IBM shipped watsonx Orchestrate for multi-agent orchestration and IBM Sovereign Core for policy-at-runtime governance. And now ServiceNow has positioned AI Control Tower as the meta-governance layer above all of them — "the AI agent of agents" — and offered it free for a year to anyone who wants it.

This convergence is not coincidence. The enterprise AI market has reached a specific inflection point: the capability layer is commoditizing, and the governance layer is becoming the primary source of vendor differentiation. Every CIO who spent 2024 evaluating AI models is now spending 2025 and 2026 evaluating AI governance platforms. The questions have changed from "which model should we use?" to "how do we see what our agents are doing, how do we control their permissions, and how do we produce an audit trail for the board?"

The Action Fabric announcement — and specifically the Anthropic launch partnership — is the most architecturally significant signal from Knowledge 2026. When Claude can route requests through ServiceNow's governance engine, the boundary between "AI assistant" and "governed enterprise actor" begins to dissolve in the right direction. The AI system is still producing the output; the governance system is ensuring that output is acting within defined permissions, is auditable, and is consistent with enterprise policies. That combination is what the regulated industries that are the most valuable enterprise AI customers actually require.

The governance wars that are starting now will be decided by two things: integration depth and trust. Integration depth means how completely the governance platform can see and control every AI system running in the enterprise. Trust means how credibly the platform can produce audit trails that satisfy the legal, compliance, and board-level accountability requirements that are crystallizing around AI governance. The platforms that win both will own a $9 billion-plus ACV market by 2030.

For enterprise leaders, the practical action item from this week is specific: before your next AI infrastructure procurement decision, ask every vendor how their system appears in ServiceNow's AI Control Tower, IBM's watsonx Orchestrate, and Microsoft's Agent 365. If they cannot answer that question, they are not production-ready for regulated enterprise deployment. The governance layer is now table stakes.

AK / Spearhead / Building AI systems, not tools