When enterprises can’t deploy your AI, you send people instead of shipping another model.
Google Cloud pulled in $24.8 billion last quarter. Alphabet is sitting on $811 billion in purchase commitments and contractual obligations as of June 30. That gap is the whole story. The spend is locked in. The demand is not. So Google is buying the one thing GPUs can’t provide: humans who know how to make the software actually work inside a company.
Google Cloud and Accenture are standing up a joint unit called the Accenture Gemini Enterprise Business Group. Google will train up to 1,000 Accenture forward-deployed engineers to sit inside customer organisations and build custom applications on Gemini Enterprise. The group reports into Accenture, not Google. The Wall Street Journal broke the arrangement first.
Forward-deployed engineers, or FDEs, are the industry’s current obsession. The theory is that enterprises aren’t short on AI models — they’re short on people who understand both a P&L and an agent framework well enough to connect the two. Every major player has now placed the same bet:
- OpenAI and Anthropic: both launched joint ventures for enterprise AI services
- Microsoft: committed $2.5 billion to its own AI deployment company
- Amazon: spun up a $1 billion FDE organisation
- Google: $750 million partner ecosystem across Capgemini, Cognizant, and Deloitte
Google also signed a multi-year deal with CVC Capital Partners to place FDEs inside the firm’s portfolio companies.
Ramp’s August data puts Google at roughly 6% of enterprise AI spending among its U.S. customers. Anthropic sits at 43.5%. OpenAI at 39.7%. Google’s response is that Ramp’s base skews away from the large strategic contracts it actually wins — Oracle, Meta, Anthropic, and ServiceNow have all signed with Google Cloud, and those deals run far deeper than model API calls.
Both things can be true. Google can win the biggest contracts and still lose the default-choice habit among everyone else. That habit is what compounds.
Accenture has its own exposure here. It ran the same play with Microsoft in March, ServiceNow in May, and SAP in June. Meanwhile firms built purely for embedded AI engineering — Ode with Anthropic, OpenAI’s The Deployment Co. — are coming for exactly the work consultancies have always owned.
If you’re buying enterprise AI, the vendor conversation just changed. You’re no longer comparing model benchmarks — you’re comparing who will staff your integration and for how long. Ask what happens when those engineers leave. And if you sell services, note that four hyperscalers just decided implementation is the margin. That’s your market now.
