CMOs, CIOs Take Note: $650B, New Agent Platforms, and a Market Share Flip
Enterprise AI Briefing: January 14 – February 9, 2026
A lot happened in enterprise AI over the past few weeks — and most of it has real implications for how organizations are thinking about infrastructure, vendors and talent.
Here's what you need to know about it all, and what to do about it.
*This is a longer issue, because there is so much to unpack, and it’s all consequential.
Top AI Stories & Developments
1. McKinsey and AWS Launched a Joint Operating Group Targeting Billion-Dollar AI Transformations
On January 22, McKinsey and Amazon Web Services launched the Amazon McKinsey Group (AMG) — a joint operating model that puts strategy, technology, and implementation into one integrated team. The stated goal is helping enterprises move from AI pilots to full-scale production, targeting transformations with $1 billion or more in client impact. McKinsey is tying fees directly to outcomes.
They've already worked with a major automaker to build an AI-enabled supply chain operating system that achieved 85% forecast accuracy and reduced inventory by 60%. When McKinsey creates a dedicated practice around a single platform partner and ties its fees to results, that tells you something about where the market is heading — and how far behind most organizations are.
2. Big Tech Just Committed $650 Billion to AI Infrastructure
Alphabet, Amazon, Meta, and Microsoft are projecting a combined $650 billion in capital expenditures this year — up roughly 60% from 2025. Nearly all of it is going to data centers, AI chips, and the infrastructure to run them. Amazon alone is spending $200 billion. To put that in perspective, 21 of the largest US industrial companies combined are projected to spend $180 billion. This is a structural reallocation of capital at a scale we haven't seen since the telecom buildout of the 1990s.
3. OpenAI Launched an Enterprise Agent Platform Called Frontier
On February 5, OpenAI introduced Frontier — a platform designed to let enterprises build, deploy, and manage AI agents that plug into existing business systems. Think of it as an orchestration layer: agents get identities, permissions, and access to institutional data, then execute workflows across tools like Salesforce, SAP, and internal databases. Uber, State Farm, Intuit, and Thermo Fisher are among the first customers.
The pitch is essentially "hire AI coworkers" — agents that onboard, learn, and improve over time the way employees do. OpenAI brought back Barret Zoph to lead the enterprise push, and CFO Sarah Friar said at Davos that enterprise currently accounts for about 40% of OpenAI's revenue with a goal to reach 50% by year-end. The urgency is real — they've watched Anthropic take share for 18 months.
4. Anthropic Now Owns 40% of the Enterprise AI Market
Anthropic's enterprise market share climbed from 32% to 40% over the past year, while OpenAI dropped from 50% to roughly 25%. In January, Anthropic landed a major deal with Allianz to deploy Claude Code across the entire organization and build custom AI agents. The company also expanded its partnership with Accenture, forming a dedicated Anthropic Business Group and training approximately 30,000 consultants on Claude.
The takeaway isn't about picking a winner — it's that the enterprise market is moving toward multi-provider strategies, and the incumbents are no longer guaranteed the default position.
5. The Per-Seat SaaS Model Is Starting to Crack
If there's one theme across every analyst report and earnings call this quarter, it's this: 2026 is the year AI stops being an experiment. Enterprises are shifting toward fine-tuned smaller models, agentic workflows that execute across systems, and standardized protocols like Anthropic's Model Context Protocol (MCP) that let agents talk to databases, APIs, and tools.
Meanwhile, Salesforce introduced an Agentic Enterprise License Agreement — essentially all-you-can-eat AI pricing. When your largest SaaS vendors start restructuring their pricing around agent consumption rather than headcount, the economics of your entire software stack are in play.
For Marketing Leaders
There are several among us, along with our core readership of CIOs and CAIOs…
The Commerce Infrastructure Is Being Rebuilt Around AI Agents
Two developments fundamentally changed how customers will discover and buy products. First, Google and Shopify launched the Universal Commerce Protocol (UCP) — an open standard, co-developed with Walmart, Target, Etsy, Wayfair, Visa, Mastercard, and over 20 additional partners, that lets AI agents handle the full shopping journey from discovery through checkout without the customer ever visiting a website. Shopify merchants can now sell directly inside Google's AI Mode in Search and the Gemini app.
Second, OpenAI announced it will begin testing ads inside ChatGPT for its 800 million weekly active users — impression-based, conversation-contextual, and structurally different from anything in the current media mix.
Meanwhile, Adobe Analytics confirmed that AI-referred traffic to retail sites grew 693% year-over-year over the holidays, and Gartner projects 25% of search volume will shift to AI by the end of 2026. If your marketing team is still optimizing exclusively for clicks and site visits, a growing share of the customer journey is already invisible to you.
What This Means for Your Organization
The consulting giants are placing their bets — and that's a signal.
McKinsey doesn't build a dedicated practice group around a platform unless they believe it's where enterprise spend is going. AMG joining Anthropic's Accenture partnership and Deloitte's Claude deployment tells you that the largest advisory firms in the world have decided AI transformation is no longer a capability area — it's the engagement.
If your organization is still treating AI as an IT initiative, you're operating with a different set of assumptions than the firms advising your competitors
The outcome-based fee structure McKinsey is using with AMG should raise the bar for how you evaluate any AI consulting or implementation partner
The infrastructure wave creates opportunity — and lock-in risk.
$650 billion in capex means the tooling and platform options available to enterprises are about to expand significantly. But every major provider is building toward an ecosystem that gets stickier over time.
The decisions you make in the next two quarters about which platforms and vendors to adopt will compound
The cost of switching later will be substantially higher than the cost of testing broadly today
Most organizations should be running multi-provider evaluations right now
Agent platforms are repositioning the entire software stack.
OpenAI and Anthropic aren't selling models anymore — they're selling the layer that sits between your people and your business systems.
If an AI agent can execute a workflow end-to-end without a human opening Salesforce, the per-seat licensing model starts to look fragile
SaaS stocks took a hit the week Frontier launched — investors see the threat clearly
Every enterprise leader should be asking: which of our current workflows are most vulnerable to agent-driven automation, and are we positioned to capture that value — or will our vendors capture it for us?
The market share shift tells you what enterprise buyers actually value.
Anthropic didn't take the lead by outspending OpenAI on marketing. They did it by being more reliable in production and more aligned with what regulated industries care about.
Production reliability, governance tooling, and integration depth are what differentiate at scale
If your team is still evaluating based on benchmark scores and demo impressions, you're behind
Selection criteria are maturing fast — your evaluation process should be too
For marketing leaders: the purchase journey is being restructured around agents.
Google's UCP and Shopify's Agentic Storefronts enable customers to discover, evaluate, and purchase without interacting with your website, your landing pages, or your conversion flow. The transaction completes inside the AI conversation.
If your product data isn't structured for agent discovery, AI agents will surface competitors by default
Attribution models built around clicks and sessions are degrading — AI-referred traffic remains largely invisible in most analytics environments
Content strategy needs to shift from driving site visits to ensuring your brand is cited and surfaced in AI-generated answers
ChatGPT advertising introduces a fundamentally different engagement model — allocate a modest test budget and monitor the early cohort, but do not redirect proven spend until measurement matures
Where to Focus in Q1–Q2
Pilot at least one agent platform (Frontier, Cowork [Megan’s personal favorite] or a competing option) against a real internal workflow, not a sandbox demo.
Audit your SaaS stack to identify workflows where agent automation could reduce or eliminate per-seat licensing costs
Pressure-test your AI governance framework — agent identity, permissions, and auditability need to be in place before deployment, not after
Negotiate AI licensing agreements with full awareness of the new consumption-based models — they favor early adopters, but lock-in risk escalates at renewal
Build internal fluency now so your team can evaluate and deploy these tools without relying entirely on vendor professional services
Marketing-specific: Establish tracking to isolate AI-referred traffic, audit product feeds for agent compatibility, and evaluate your brand's visibility in AI-generated answers across ChatGPT, Gemini, and Perplexity
Bottom Line: The enterprise AI market is consolidating around agent platforms and infrastructure plays — and the commerce layer is moving just as fast. The organizations that move from pilots to production in the first half of 2026 will have a measurable structural advantage. The ones still running exploration committees will be playing catch-up by Q3.
One more thing. We're putting together our Q1 Enterprise AI Readiness Report — a benchmark study of how organizations are actually adopting, deploying, and measuring AI. If you've been wondering how your organization compares to peers, this is how you find out. We're selecting a small group of enterprise leaders to participate, and participants get the results before anyone else. Reply with "Invite requested" and we'll send you the details.
