When Your Vendor Becomes Your Competitor
Anthropic, Microsoft and OpenAI have each built implementation arms in the last ten weeks, backing more than $8 billion combined. OpenAI's newest product goes further: it can only be deployed through approved partners OpenAI itself selects. The vendor isn't just competing for delivery work anymore. It's also deciding who gets in the room.
- market structure
- AI vendors
- implementation
- private equity
- competitive dynamics
In May, Anthropic put a name on something it had been building quietly with Blackstone and Hellman & Friedman. The name launched in July: Ode with Anthropic, a standalone services firm with Anthropic's own engineering and partnership teams embedded inside it, built to bring Claude into mid-size companies across healthcare, manufacturing, financial services, retail, real estate and infrastructure. Reported backing is $1.5 billion, from a consortium that also includes Goldman Sachs, General Atlantic, Apollo, GIC and Sequoia.
That same window, Microsoft confirmed it was committing $2.5 billion to something it calls Frontier Company: roughly 6,000 of its own engineers, redeployed from existing forward-deployed and consulting teams, tasked with building and running AI systems inside customer organizations rather than licensing the software and leaving. And back in June, OpenAI had already priced its own version at more than $4 billion: the Deployment Company, built around the Tomoro acquisition's 150 engineers, backed by TPG, Advent, Bain Capital, Brookfield, and a partner list that includes Bain & Company, Capgemini and McKinsey.
Three model labs. Three implementation arms. More than eight billion dollars of disclosed capital inside about ten weeks. I wrote about the OpenAI move alone back in August, in OpenAI Spent $4 Billion on the Implementation Problem. Read the three side by side and the interesting part isn't any single one of them. It's that all three made the same bet, on close to the same timeline, about where the money in enterprise AI actually sits.
Delivery capacity was the easy part
A model vendor building a delivery arm and competing for the same engagements an independent agency already bids on is a straightforward story, if an expensive one. It's what OpenAI, Anthropic and Microsoft have each just done.
What changes the shape of it is OpenAI Presence, the enterprise agent platform OpenAI launched the same month the Deployment Company news finished circulating. Presence isn't self-service. OpenAI's own description is specific: it's available through "a limited general availability program," and deployments are led by OpenAI's own Forward Deployed Engineers "and select global systems integrators." No public list of who qualifies. No published certification path. The company that builds the model is now also deciding, deal by deal, which outside firms are allowed to touch its newest product at all.
That's a different relationship than competition. A competitor bids against you for the same client. A gatekeeper decides whether you're in the room to begin with. OpenAI is doing both, for the same product, in the same quarter it opened a $4 billion delivery arm of its own.
The consultants funding their own replacement
Here's the detail worth sitting with. OpenAI's Deployment Company counts Bain & Company, Capgemini and McKinsey among its financial backers. Three of the firms whose consulting revenue this structure is built to compress have bought a stake in the entity built to do the compressing.
That's not hypocrisy. It's a hedge, and a rational one: if implementation work is migrating toward vendor-owned delivery arms with frontier-model access no independent firm can match, owning a piece of the arm doing the migrating beats standing outside it. The uncomfortable read is what it says about where the smartest people in the industry think the next several years of margin actually land. Not with the firms that write the strategy deck. With whoever's engineers are already inside the account when the model improves again.
What none of this touches
None of the three arms solves the problem that made implementation the bottleneck in the first place. A forward-deployed engineer, whether OpenAI's, Anthropic's or Microsoft's, still needs someone to tell them what a company's workflows actually are before there's anything to redesign. Capital buys the engineer. It doesn't buy the description of the work, and that description doesn't get cheaper because the entity holding it has more money behind it.
That's the entire advantage available to a company below the size these programs are built for. It was never going to win a bidding war against $2.5 billion, and it doesn't need to. The question these three launches actually raise for a mid-market buyer isn't which vendor's delivery arm to hire, because within a year or two all three will likely be offering some version of the same motion at some version of the same price. It's whether, before any of them shows up, you can hand any engineer from any of the three a description of your own operation precise enough that they could start on day one instead of spending the engagement's first month figuring out what you do.
Three vendors just spent a combined eight billion dollars proving they believe implementation is where the value sits. None of them has said a word about the one input that decides whose implementation wins.
If you want that question answered for your specific situation, the Forge Playbook does it. Answer a few questions about your business and we'll put together a tailored outline of which workflows are worth automating and what a realistic budget looks like for each. Free, no obligation, takes about three minutes.
