Enterprise AI's infrastructure moment: agents, gateways, and the battle for the control plane

OpenAI launched ChatGPT Work on July 9, built on its new GPT-5.6 model and aimed squarely at automating complex, multi-step workplace tasks. The product puts OpenAI in direct competition with Microsoft and Google on enterprise turf both companies have spent years building, according to Forbes reporter Madhulika Pathak. The timing is deliberate: agentic AI, where software acts autonomously rather than waiting for a human prompt, is rapidly becoming the primary battleground for enterprise platform dollars.

Agent gateways: the infrastructure layer operators can't ignore

The push into autonomous agents is creating an urgent secondary problem for IT and security teams: how do you govern something that governs itself? Forbes contributor Janakiram MSV reported this week that agent gateways are emerging as the answer, functioning as a control plane that sits between enterprise environments and AI agents to enforce policy, log actions, and manage access.

The vendor lineup moving into this space is not small. Nutanix, Palo Alto Networks, and AWS are among the major players actively building agent gateway capabilities, according to Forbes. For a CIO or CISO evaluating an agentic AI rollout, the gateway layer is no longer optional infrastructure; it is the mechanism that makes an agent deployment auditable and compliant.

The pattern mirrors what happened with API gateways a decade ago, where raw connectivity needed a governance wrapper before enterprises would trust it at scale. Agent gateways are filling the same role for autonomous systems, and the speed at which major cloud and security vendors are entering the space suggests procurement timelines will compress quickly.

The 130% valuation gap and what it tells procurement teams

While the infrastructure debate plays out, a McKinsey study is reshaping how CFOs and product leaders frame AI investment. Forbes contributor Jodie Cook reported the finding this week: companies that embed AI directly into their products carry valuations approximately 130% higher than those applying AI only to internal operations. The gap is not marginal.

For enterprise operators, the implication is concrete. AI spend directed purely at back-office productivity, workflow automation, or cost reduction delivers a measurably smaller return than spend that reaches the product or customer experience. That distinction matters enormously for how procurement directors prioritize vendor evaluations and how IT leaders justify platform choices to CFOs who are, separately, scrutinizing every AI line item.

The Wall Street Journal reported this week that Amazon recently joined a wave of large U.S. companies raising capital through bond markets specifically to fund AI infrastructure investment. The scale of that capital commitment, from hyperscalers to enterprise software vendors alike, suggests the infrastructure build-out underpinning tools like ChatGPT Work and agent gateways is still accelerating, not plateauing.

Three decisions converging for enterprise operators

The developments this week are not isolated. ChatGPT Work raises an immediate evaluation question for any organization running Microsoft 365 Copilot or Google Workspace AI features: is the incumbent platform still the right agent layer, or does a purpose-built agentic tool now warrant a side-by-side comparison? The competitive pressure on enterprise AI vendors is real and it runs in both directions.

The agent gateway story is a procurement and architecture question in parallel. Organizations already deploying pilots with autonomous agents need a governance answer before they scale. Without a gateway layer, audit trails are incomplete, security policy enforcement is manual, and the risk surface grows with every agent added.

The valuation data from McKinsey, meanwhile, gives enterprise technology teams a sharper internal argument for moving AI investment toward products and customer-facing services. The next concrete marker to watch: how quickly Microsoft, Google, and AWS respond to ChatGPT Work's launch with their own product-level agent announcements, and whether agent gateway standards begin to consolidate around one of the major cloud platforms or remain fragmented across vendors.

What this means for your team

• Evaluate ChatGPT Work against your current enterprise AI platform on task complexity and multi-step automation capability, not just per-seat cost.
• Audit your current agentic AI pilots for governance gaps: if you cannot produce a full action log for every agent decision, a gateway layer should be on the near-term roadmap.
• Reframe internal AI investment proposals around product-level or customer-facing deployment to align with the McKinsey valuation premium data when presenting to finance leadership.
• Track vendor gateway announcements from Nutanix, Palo Alto Networks, and AWS over the next 60 days; consolidation in this space will affect which platforms are easiest to govern at scale.