The firm PE sponsors call when they do not want the model vendor in the room.

The common reading of these announcements is that private equity has decided to buy its own AI implementation capacity at the portfolio level. That reading is correct as far as it goes. It is also incomplete, because both vehicles are, by construction, narrower than the work they claim to address.

The shape of the JVs

Both are single-vendor. The Anthropic vehicle exists to embed Claude into portfolio company workflows; the OpenAI vehicle exists to distribute Frontier, the enterprise product line. Both rely on the global systems integrators for delivery. Anthropic's implementation layer is the Claude Partner Network, a one-hundred-million-dollar program anchored by Accenture, Deloitte, Cognizant, and Infosys, announced in March 2026.^[3] OpenAI has been publicly recruiting consultancies to plug its own implementation gaps.^[4] In both cases, the execution layer is the same rate-card-heavy SI machinery that PE operating partners already complain is slow.

Both put a sponsor on the cap table. Blackstone, Hellman & Friedman, and General Atlantic hold equity in the Anthropic vehicle. TPG, Advent, Bain, and Brookfield hold preferred equity in the OpenAI vehicle. This is the structural detail that matters most. It is what makes the JVs efficient — one conversation at the operating partner level can replace a hundred enterprise sales cycles at the portco level — and it is also what forecloses a set of engagements neither JV can accept without prejudice.

What gets left on the table

The residual opportunity divides into five gaps.

The first is model neutrality. A single-vendor JV cannot, by definition, recommend the competing vendor's model. But the decision about which model wins a use case is increasingly a decision about the use case, not about the firm. A portfolio company that needs Gemini for multimodal, Llama for on-prem inference, and Claude Opus for long-context coding needs a firm that is allowed to recommend all three. Neither JV can be that firm. An independent advisor can.

The second is sponsor neutrality. A Blackstone-backed integrator flowing resources into an Apollo or KKR portfolio is politically awkward in a way that is not easy to paper over. The discomfort is sharpest for the sponsors that are not on either JV's partner list — the long tail of five-to-fifty-billion-dollar funds that will be treated, at best, as a secondary market. An independent advisor has no sponsor on the cap table and no conflict to disclose.

The third is speed. PE operating partners are chronically frustrated by six-month Big Four discovery phases. The reason is mechanical: SI economics reward utilization, and utilization rewards length. A firm that is not optimizing for utilization can run a four-week diagnostic and a twelve-week implementation against a fixed fee, and can do so honestly.

The fourth is industry specialization. PE owns deeply specific industries — vertical SaaS, industrial distribution, physician practice management, specialty finance, specialty manufacturing — where generic large-language-model playbooks fail. JV delivery layers are horizontal by design. An independent firm can choose a narrow set of verticals and go deep.

The fifth is post-deployment governance. Neither JV has publicly addressed what happens after go-live. Model drift, evaluation pipelines, sponsor reporting, value attribution — these are durable, tenure-heavy obligations that a utilization-driven SI is structurally unsuited to. A forward-deployed engineer who stays through the first earn-out cycle and translates technical outcomes into operating-committee language is a product the JVs cannot credibly offer.

Why the window matters

The JVs are still in talks. As of April 2026, no final close has been reported, no launch date announced, and no leadership named in the public record.^[1][2] The market is watching, operating partners are asking what this will mean for their portfolios, and the answer — for the many firms not on either partner list, and for the many use cases that do not fit a single-vendor shape — is that the field is open.

The window is approximately twenty-four months. Once the JVs launch, name their first dozen portco engagements, and settle into their cadence, the category will harden. The firm that wants to be the independent option needs to be in the market before the category is settled, with a visible thesis, a credible engagement model, and the beginnings of a reference list.

The posture of the firm

The firm the market needs is small, expensive, and uninterested in competing on price. A twenty-percent discount does not win a deal in this category; being model-neutral and sponsor-neutral does. Rates should sit at parity with the JV rate cards, not below them. Cheap reads as junior, and a portco CEO does not want to defend a junior recommendation to the sponsor.

Services revenue should stand on its own P&L. The Claude Enterprise subscription that sits underneath a JV engagement is the thing an independent firm cannot replicate and should not try to. The implication is gross margins earned on the work itself, which in turn implies no Accenture-style bench, a deliberately small senior-heavy team, and the discipline to decline engagements that cannot be staffed honestly.

The ambition should be narrow and legible. The firm is not an AI consultancy, because that word is reserved for a different market. It is a forward-deployed engineering firm, the way Palantir was a forward-deployed engineering firm before the phrase entered the vernacular. It takes its name from the writing system devised by Fëanor in Tolkien's legendarium — letters designed to encode any language with precision. The work is the same: encoding the operating knowledge of a business into systems that endure beyond any one engagement.

The sentence that settles it

The firm PE sponsors call when they do not want the model vendor in the room is a different firm. It has to exist because the JVs are structurally barred from being it. It has to be independent, model-neutral, and sponsor-neutral. It has to run a fixed-fee diagnostic that an operating partner can greenlight without an investment committee. It has to ship production systems in a quarter, not in a year. And it has to stay through the first earn-out cycle, because that is when the operating committee finally reads the numbers.

That firm is Tengwar. The reason the firm exists is on this page.