OpenAI and Anthropic are not becoming consultancies. They are trying to tackle the AI last-mile challenge

Anthropic and OpenAI have thrown down the gauntlet by announcing audacious moves backed by leading PE companies to create AI services companies to scale deployments and ultimately solve last-mile challenges. But to whom? To the IT services community to rip up their go-to-market playbooks or to the investment community to raise the pre-IPO stakes and justify the lofty valuations? While tempting to add clickbait by proclaiming that management consultants and GSIs are toast, answers to these questions are as complex as the topic itself. And we don’t know many details as yet. Anthropic’s announcement was refreshingly vague, while OpenAI’s move (at the point of writing) is based on rumors. The repercussions of those moves could be profound: 1. new levers to accelerate enterprise adoption of AI. 2. disruption of the IT Service provider landscape and 3. a new playbook for the mid-market. Let’s look at the announcements in more detail.

AI last-mile challenge meets the mid-market conundrum

Let’s start with what we know. Anthropic stated that it plans to create an AI services company that will work with mid-sized companies across sectors to bring Claude into their most important operations. Blackstone, Hellman & Friedman, and Goldman Sachs back the move. Beyond those founding partners, a consortium of alternative asset managers, including General Atlantic, Leonard Green, Apollo Global Management, GIC, and Sequoia Capital, is supporting. Anthropic is said to have collected $1.5 billion for this move.

Anthropic cites that “enterprise demand for Claude is significantly outpacing any single delivery model.”  But the intent appears to go much deeper than “just” demand. The strategic goal is to create a company that will offer FDE capabilities to the mid-market. One facet of context is OpenAI’s Frontier initiative, which brings together McKinsey, BCG, Accenture, and Capgemini to work alongside OpenAI’s FDE team to scale AI deployments. The other facet is to have global partners to serve the mid-market. This is typically the domain of broadliners and VARs. System integrators have tried to crack this segment, but a lack of skills and the cost of sales have been difficult to overcome. Accenture is just giving it another go, creating a new mid-market unit under the guidance of veteran Kishore Durg.

If rumors are corroborated, OpenAI has the ambition to launch The Deployment Company, its main backers include TPG Inc, Brookfield Asset Management, Advent, and Bain Capital, while Anthropic draws upon Blackstone, Hellman & Friedman, and Goldman Sachs. OpenAI’s investment vehicle has raised $4 billion from investors to value the new company at $10 billion, while Anthropic is said to have collected $1.5 billion. The only difference is that Anthropic places a strong emphasis on the mid-market.

Palantir’s FDE playbook might be a blueprint, but AI needs a reinvention of that playbook

It is tempting to suggest that OpenAI and Anthropic are throwing the Palantir playbook at the sluggish pace of AI adoption in the enterprise. Yet the real context appears to be that there is a growing realization across the industry that enterprise adoption doesn’t stall because of insufficient models, but because of a lack of operating model reimagination and the resolution of last-mile issues. Put another way, currently, in most cases, standardized AI components are consumed through ISV platforms or GSI marketplaces just to hit the harsh realities of technology, process, and cultural debt. Palantir has spent more than a decade building integration layers, data ontologies, deployment methodologies, and an embedded operating model. Thus, it offers much more than a contingent of FDEs accelerating deployments. What is most compelling is the ability to trace process steps, which enables work in highly regulated sectors such as the Public Sector. If anything, for Anthropic and OpenAI to be successful, it requires a new playbook that blends the notion of FDE teams with operating model reimagination and the solving of last-mile issues. But building this will take time.  

OpenAI is trying to have its cake and eat it

That this playbook must go beyond Palantir’s version becomes clear when analyzing OpenAI’s two initiatives, Frontier and The Development Company. Frontier is about platform control, aiming to become the agentic operating layer. Crucially, the launch partners are the high-value tip of the partner iceberg. McKinsey, BCG, Accenture, and Capgemini are tasked to work alongside OpenAI’s FDE team to scale AI deployments. Thus, FDEs are just one component of a much broader strategy. Just as with Anthropic, The Development Company seeks to leverage its PE backers’ portfolios as a distribution channel. The strategic goal appears to be industrialized implementation, possibly through acquired AI services firms and the poaching of teams. The new vehicle partially bypasses the partner channel. It creates a more controlled deployment route, especially when PE owners want fast productivity gains across portfolio companies. That is not classic enterprise consulting; it is closer to a deployment factory for AI value capture.

This ain’t no Avanade

It is not easy to find a comparison for the moves a product-led company makes to become a services-led organization. The closest might be IBM’s 1990s turnaround, which moved it from selling boxes to integrating, operating, and transforming enterprise IT.  Conversely, the creation of Avanade by Microsoft and Accenture was more about amplification of an ecosystem than taking direct ownership of deployments. More recently, we saw ISVs like ServiceNow trying to find answers by taking stakes in partners like Plat4mation, explicitly to address the mid-market conundrum – only to run into the perennial issues of cost of sales and relevant talent. Thus, PAC expects the battle lines to be drawn around enterprise adoption of AI, with the mid-market remaining a sideshow at best.

Bottom line: The last mile is the battle line for value realization and stack control

The winners will not be those with the most AI certifications. They will be those who can prove they own the last mile of value capture: process redesign, data context, controls, adoption, operating model change, and measurable business outcomes. This used to be the domain of the Big 4 and the GSIs, but the frontier labs are threatening to disrupt the status quo.

Both moves are partner-friendly on the surface, but there is a clear competitive tension beneath the surface. The positive spin is about providing more talent to accelerate AI adoption; the more cynical reading is about the race to control value realization and ultimately owning the control points in the AI stack. This would explain the difference to an Avanade-style provider-backed approach.

Rather than focusing on the mid-market, PAC believes that Anthropic will focus on the enterprise segment while opportunistically leveraging some of the PE portfolio companies as distribution channels. However, to succeed in the mid-market, OpenAI and Anthropic would have to write a completely new playbook, not just poaching teams or acquiring assets. Yet, focusing on the mid-market might distract from a conceivably bigger prize of achieving a successful IPO.

In summary, value realization in AI will not sit with whoever builds the best model, but with whoever governs the last mile: workflow orchestration, evaluation, autonomy, and business accountability. That is why hyperscalers, workflow platforms, frontier labs, and service providers are all converging on the same control points. OpenAI and Anthropic have upped the ante in this strategic battle. Their moves will put these issues under the spotlight, but solving them will take considerable time and more than PE money. And no, both companies won’t become service companies.