The Rise of AI Services at AI Labs
Selling AI and implementing it within enterprise environments have emerged as distinct businesses, and the companies developing AI models and platforms are increasingly taking a direct role in deployment and integration.
Over this spring and summer, several major enterprise technology companies independently moved toward expanding their AI deployment capabilities through embedded engineering and services organizations. Salesforce launched its certified partner network in mid-April, followed by Google Cloud’s $750 million commitment at Cloud Next. Anthropic announced a services company (Ode) with Blackstone, Hellman & Friedman, and Goldman Sachs in early May, while OpenAI followed with a $4 billion venture led by TPG (OpenAI Deployment Company). Databricks formalized its delivery organization in June, AWS committed $1 billion to a dedicated unit, and Microsoft announced a new operating business with 6,000 employees and $2.5 billion in funding. Snowflake has also developed similar capabilities without a comparable public announcement.
The broader pattern across companies with different business models is more significant than any individual investment. Model providers, cloud platforms, data platforms, and enterprise software vendors are reaching a similar conclusion: whether the underlying product is models, compute, data, or workflows, customers still require significant integration and implementation work to deploy AI effectively (https://elaxtra.com/insights/the-emergence-of-the-agentic-fde). This work has traditionally been performed by consulting firms and systems integrators, while the increasing involvement of technology vendors suggests that they see strategic value in owning or directly influencing this layer.
These initiatives also differ materially in how the services capability is structured. AWS, Microsoft, Databricks, and Snowflake place their own engineers within customer environments, funding delivery internally and maintaining direct customer relationships. Anthropic and OpenAI have taken a model closer to a standalone consultancy, with external capital supporting dedicated services organizations that can develop their own client base. Google Cloud has instead focused on embedding engineers through existing systems integrators, while Salesforce combines a smaller internal team with a broader certified partner network designed to scale delivery.
The distinction between who owns the customer relationship and who funds delivery has important economic implications. Vendors that keep engineers in-house are effectively betting that the delivery layer provides enough strategic value to justify the associated fixed costs and enable them to capture services margin directly (https://elaxtra.com/insights/the-improved-economics-of-agentic-services). Vendors that rely on partner-embedded engineers are instead leveraging firms that already hold customer relationships and manage the costs associated with delivery capacity. The relative attractiveness of each model will depend on whether embedded engineering creates sufficient integration, differentiation, or switching costs to justify the investment.
The common conclusion is that the model itself is increasingly only one component of the product; the deployed system and the outcomes it enables are becoming more important (https://elaxtra.com/insights/pricing-services-from-time-to-outcomes). Enterprise AI adoption can be constrained not only by model capability but also by the engineering and integration capacity required to move from pilots to production. As a result, companies providing infrastructure for enterprise AI are increasingly participating in deployment and implementation. The extent to which they retain or outsource that activity will shape their economics over the coming years.
Elaxtra Advisors is an M&A and value-creation advisory firm that assists institutional investors, private equity-owned platforms, and strategic acquirers invest and create value in worldwide technology services companies. Please contact us to explore potential partnerships.