Hg and Anthropic Are Running OpenAI's Playbook for Embedding AI in Portfolio Companies
A new Hg-Anthropic venture to embed AI engineers inside client companies is the second version of a model OpenAI built with TPG, Advent, Bain Capital and Brookfield five months earlier.
LONDON, November 3, 2026. Hg and Anthropic are forming a new company. Its job: embed AI engineers inside client businesses, starting with Hg's own portfolio. London and Berlin will host the first offices. Launch is targeted by year-end. The Financial Times broke the story October 23. Neither firm has confirmed it on the record.
The structure splits the capital two ways. Anthropic funds its side from its own balance sheet. Hg invests through its managed funds, including the listed HgCapital Trust. Hg already runs an internal AI deployment team of about 150 people across its portfolio. The new venture sits alongside that team, not inside it. People familiar with the plans told the FT one number. A valuation target above $10 billion, within five years.
OpenAI built something similar in May. TPG led a roughly $4 billion raise, with Advent, Bain Capital and Brookfield joining in. The venture is called the OpenAI Deployment Company. It runs on Tomoro, an applied-AI consultancy OpenAI acquired, along with about 150 forward-deployed engineers. Bain & Company, Capgemini and McKinsey signed on as integration partners. The model in both cases is the same. A frontier lab supplies the technology and some of the talent. PE capital supplies the balance sheet, plus the portfolio companies to practice on.
Hg and Anthropic's version is smaller and more concentrated. One lab, one firm, one portfolio to start. OpenAI spread the model and the risk across three PE firms and three consulting firms. Both are betting on the same thing. The model is rarely the hard part of enterprise AI. Running it inside a specific company's workflows is harder. Forward-deployed engineers, not software licenses, do that work.
The timing isn't an accident. PE fundraising is splitting in two. PrivateEquityList counted 87 new private equity and venture capital funds launched in September, worth $98.5 billion combined. Fourteen funds of $1 billion or more took in 86% of that total. Thirty-nine of the 87 funds, nearly half, carry an AI tag. Capital is concentrating in firms that can point to an AI thesis.
Not every firm is reaching for a lab partnership to prove it. Blackstone built its own unit, BXN1. It says it has deployed more than $150 billion into AI infrastructure, including data centers. The firm also touts more than 5.8 trillion data points under its management. Goldman Sachs Alternatives is making direct bets instead. It backed Fieldguide, an AI audit and accounting platform. It also backed BLP Digital, a Swiss invoicing-automation firm.
Three different paths, one shared read on where the money is going. The test for all of them is the same. Does embedding AI inside a portfolio company's operations produce returns a generic software license couldn't? Hg and Anthropic will find out first, starting with Hg's own portfolio.
