Summary

• The Financial Times reports that Anthropic and Hg have agreed to establish a European AI consultancy expected to launch by the end of 2026.
• The proposed venture would place engineers and consultants inside businesses, including companies owned by Hg investment funds.
• The financial terms, ownership structure and leadership have not been disclosed, while the partners’ existing AI collaboration is already operational.

Anthropic and European software investor Hg reportedly plan to establish an independent AI consultancy with operations in London and Berlin, extending their existing partnership into a business that would place specialist engineers inside companies developing AI-enabled products.

The proposed venture is expected to launch before the end of 2026, according to reporting by the Financial Times on 23 October. It would work with European corporate customers, including software businesses within Hg’s investment portfolio, while operating separately from the investment firm’s established internal AI teams.

Anthropic is reportedly providing capital from its balance sheet, while Hg would invest through funds under its management, including HgCapital Trust. Neither the investment amount nor the proposed ownership split has been disclosed, and the companies have not independently confirmed the new business through an official launch announcement.

The reported agreement builds on a commercial relationship that Hg has described publicly. On 21 September, the investor announced an expansion of its work with Anthropic to develop agentic AI capabilities across its software portfolio, including collaboration between Anthropic and the Hg Catalyst product incubator.

Although the existing programme is confirmed, it is distinct from the separately capitalised consultancy described in the October report. Treating Hg’s portfolio activity as proof that the new company is already operating would overstate the available evidence.

A partnership with an established delivery operation

Hg began working with Anthropic in early 2024 and has used that relationship for product development, training and engineering projects among companies in which its funds invest. The portfolio contains businesses providing specialised software for financial administration, payroll, tax, healthcare and other information-intensive activities.

Those products frequently hold important customer records and perform processes governed by detailed business rules. Introducing AI functionality can involve more work than connecting a model to a user interface, particularly where automated outputs influence financial records, regulated activities or decisions affecting customers.

Hg Catalyst and the investment firm’s broader engineering capability work with portfolio companies on such tasks. The investor says more than 1,600 AI projects are live across its businesses, representing over $260 million of budgeted impact on earnings before interest, tax, depreciation and amortisation.

That figure is a company-reported estimate of expected financial effect across existing portfolio initiatives. It is not a verified amount of realised savings, revenue from AI services or income attributable to the proposed consultancy.

The Financial Times reports that Hg already has approximately 150 AI specialists involved in supporting portfolio companies. A new independent business would therefore enter a market in which its proposed investors have practical implementation experience, rather than building every capability from the beginning.

However, the relationship between the proposed consultancy and Hg’s internal specialists remains unclear. Customers and investors would need to understand how work is allocated, which organisation contracts with clients and what resources are shared.

Enterprise AI brings systems integration work

Many companies testing AI models can demonstrate useful responses to selected questions. Deploying the same technology reliably in an established business process requires additional engineering, information governance and operational controls.

A software provider serving accountants, for example, might want an AI assistant to examine transactions or prepare draft explanations. Turning that capability into a commercial product requires connections to financial records, permission checks, suitable tests and procedures for handling outputs that cannot be trusted.

Businesses operating regulated services also need to determine whether automated activity can be reconstructed and reviewed. A response that appears plausible may be insufficient if it cannot be connected to the records supporting the underlying conclusion.

Embedded engineering teams can help adapt an organisation’s existing applications, data structures and working methods. Their tasks may include designing interfaces, testing model behaviour, preparing deployment environments and establishing monitoring arrangements once a product begins operating.

Such services can be commercially valuable where a customer lacks specialist developers or needs additional capacity for a defined project. They also carry ongoing obligations concerning maintenance, model changes and the performance of applications after the initial delivery.

Customers must decide whether an external specialist offers enough additional expertise to justify consulting fees, model usage charges and the internal resources needed to support the project.

Model providers deepen their commercial relationships

The reported European venture follows other arrangements through which AI model developers have become more closely involved in delivering business applications.

Anthropic is associated with Ode, a US-focused AI services company backed by investment firms, while OpenAI has also pursued investment partnerships linked to enterprise implementation. Such businesses operate alongside established technology consultancies and systems integrators that already provide software development and organisational change services.

For a model supplier, participation in deployment services could create closer relationships with customers using its technology. It may also provide information about the difficulties businesses encounter when introducing models into production systems.

Those commercial advantages would need to be balanced against existing relationships with external implementation partners. Large AI providers depend on consultancies, cloud companies and independent developers to reach customers whose requirements they cannot support directly.

The model supported by the proposed Hg–Anthropic venture has not been disclosed in sufficient detail to establish whether it would exclusively implement Anthropic products or work with technology from competing suppliers.

An exclusive arrangement could create a closer relationship with one model platform, while a wider approach might be more attractive to customers seeking to compare products or reduce dependence on a single supplier. Neither outcome should be assumed before the commercial terms become public.

Investment ownership adds another consideration

Hg’s funds own software businesses that could become customers of the proposed consultancy. Such relationships may provide a source of early projects, particularly where companies are already participating in the investor’s AI development programme.

They also create questions about commercial governance. Portfolio businesses need to evaluate the cost and suitability of services provided by an organisation in which their financial sponsor also holds an interest.

That assessment is especially important when an AI implementation involves substantial changes to a commercial product. Decisions about model selection, data access and maintenance can influence future operating costs and the company’s ability to work with alternative providers.

The reported venture’s independence would require clearer information about governance, management, customer contracts and the allocation of responsibilities between the consultancy, Anthropic and Hg.

No management appointments, launch date or confirmed customer engagements for the proposed company have been announced. Its anticipated London and Berlin hubs are part of the reported plan rather than evidence that staffed offices and operations have already been established.

Hg’s existing partnership with Anthropic provides a credible commercial background, but it does not resolve those outstanding details. Formal confirmation of the new company and its investment terms will be needed before its scale or competitive position can be properly assessed.