Why Anthropic Stopped Competing for Software Budgets
TCS and DXC aren't reselling Claude. They're using it to make the workflows they already own harder to ever take back.
A TCS subsidiary called Diligenta administers more than 21 million life and pensions policies for insurers including Scottish Widows and Phoenix Group. No procurement committee approved that arrangement this year. The contracts predate the current AI cycle by a decade.
That detail matters more than any partnership press release, because it explains why Anthropic’s recent expansion into TCS, DXC, Accenture, Deloitte, Cognizant, KPMG, and PwC has been misread. Most coverage treats these deals as channel-building: Anthropic found bigger sales partners. The more consequential story is about who controls the workflow Claude enters, and what that means for every software company that assumed AI competition would be fought model against model.
For strategy leaders inside enterprise software, distribution executives at AI labs, and CFOs evaluating renewal risk on core systems, this is the conversation that matters now: ownership of the workflow is becoming more valuable than ownership of the application that sits on top of it.
The Workflow Was Already Spoken For
The prevailing assumption in enterprise software has been straightforward for thirty years: vendors compete for ownership of a workflow, customers run a procurement process, and switching costs protect whoever wins it. That assumption breaks when the workflow already belongs to someone else before any AI vendor shows up.
TCS does not approach an insurer as a new entrant pitching a budget line. It already operates the claims handling, policy administration, and customer service infrastructure those insurers depend on. When Claude is introduced into that environment, it enters through an existing operating relationship rather than a competitive RFP. There is no vendor evaluation, because the evaluation already happened years ago when the outsourcing contract was signed.
DXC’s OASIS platform makes the mechanism concrete. Claude generated more than 95 percent of the code behind it, and DXC has deployed it across more than 50 client environments. The commercially important detail is not the code-generation statistic. It is that OASIS is bundled into managed-services contracts rather than sold as standalone software, which means clients receive it without a separate vendor approval cycle.
The conventional view holds that Anthropic is winning enterprise distribution. The more accurate view is that distribution was never up for grabs in the way the software industry assumed. It was already owned by the firms operating the workflow, and AI just made that ownership more monetizable.
A Services Industry Defending Its Existence
It is tempting to read these partnerships as confidence. The more honest read is urgency. The top twenty global systems integrators generated nearly $450 billion in revenue in 2024, yet their share of total IT spending has been eroding for years as clients shift budget toward cloud providers, boutique consultancies, and in-house engineering teams. Generative AI now threatens the labor-intensive activities, documentation, testing, migration, requirements gathering, that helped these firms scale in the first place.
That is the context the partnership headlines tend to omit. Cognizant has been explicit about repositioning away from traditional systems integration toward AI-enabled delivery. Accenture has committed substantial resources to certifying practitioners on Anthropic’s models. The Claude Partner Network drew more than 40,000 applications and produced over 10,000 certifications within months, and the top tier requires sustained deployment activity across multiple geographies, not a one-time training session.
Anthropic’s $1.5 billion AI services venture, backed by Blackstone, Goldman Sachs, and Hellman & Friedman, reinforces the same thesis. Goldman’s Marc Nachmann framed the venture around expanding access to forward-deployed engineering talent, which is a tell. The scarce resource was never model access. Most enterprises can already reach a frontier model. Almost none have teams capable of integrating it into regulated, production environments without breaking something.
Read that way, the certification push is not ecosystem marketing. It is Anthropic building a distributed deployment workforce inside organizations that are simultaneously racing to prove their labor model still matters in an AI economy. Both sides have reasons to move fast, which is why the partnerships have scaled faster than typical enterprise channel programs.
Where Renewal Leverage Goes Next
The real test of this shift will not show up in partnership announcements. It will show up at renewal, when