OpenAI has agreed to acquire Northslope, an applied-AI company whose founders came from Palantir. The deal is the second purchase by the OpenAI Deployment Company, a majority-owned subsidiary OpenAI launched in May with more than $4 billion earmarked specifically for acquisitions like this one.
The first acquisition was Tomoro, an Edinburgh-based AI consulting firm with around 150 engineers who had already deployed AI systems for clients including Tesco, Virgin Atlantic, and Fidelity International. Northslope adds hundreds more of what the industry is calling forward-deployed engineers: people who move into a client's offices, learn how the business runs, and build AI systems around its actual processes rather than selling a generic product.
Northslope specialised in building AI applications on top of Palantir's own software stack, meaning it enters with existing relationships inside Palantir's enterprise customer base. OpenAI is not just buying headcount. It is buying a proven method, a specific client network, and institutional knowledge that took years to build.
The broader picture is striking. In the span of a few weeks, Microsoft committed $2.5 billion and 6,000 staff to its own version of this model. Amazon put $1 billion behind a similar initiative. Anthropic launched a $1.5 billion joint venture backed by Goldman Sachs, Blackstone, and Hellman and Friedman, focused on embedding engineers inside mid-sized companies through the investment firms' own portfolio networks. The combined spend from these four players alone exceeded $8 billion.
The reason is plain. AI models from different companies are getting more similar by the month, prices are falling, and selling raw model access is becoming a lower-margin business. The money is in making the technology produce real outcomes inside a specific company's operations. As Goldman Sachs' global head of asset and wealth management said when discussing the Anthropic venture: "having the model alone doesn't change your workflows or how you operate."
All of this is landing on the traditional consulting industry at a difficult moment. Accenture lost nearly 20% of its stock value in a single day in June 2026, its worst-ever trading session, after reporting softer revenue guidance. Bloomberg Intelligence linked the sell-off directly to AI reducing demand for conventional consulting and managed services.
The irony is real: consulting firms built revenue for decades by helping clients adopt new technology. Now the technology companies are hiring their own consultants and going direct.
For any business currently considering an AI project, this shift has a practical side. OpenAI's deployment arm will almost certainly focus its limited engineering capacity on its largest and most complex clients first. The same goes for Microsoft and Amazon's versions of this. Mid-market and smaller operators are more likely to find themselves served through the private equity networks backing these ventures, particularly Anthropic's, which was explicitly structured to reach companies through its investors' portfolios.
There is also a lock-in question worth taking seriously. Deployments built by OpenAI's engineers, using OpenAI's models and tooling, will over time make switching to a competitor harder. Microsoft's CEO Satya Nadella has publicly argued that customers should be able to swap one AI model for another without losing everything they have built. Whether any of these embedded-engineer programmes actually deliver that in practice, as opposed to promising it, is the right question to ask before signing.
The regulatory approval required for both Northslope and Tomoro deals is still pending. Until deals close, the engineers are not yet formally part of OpenAI's organisation. But the direction is clear, and the pace of acquisitions suggests the $4 billion will not sit idle for long.