Google Cloud hires Accenture's engineers to turn Gemini spending into revenue
A new Accenture unit will put 1,000 engineers inside companies to deploy Gemini Enterprise. The deal shows that Big Tech's AI returns now depend on consulting labour as much as on models.
Google Cloud and Accenture on Tuesday, September 8, launched the Accenture Gemini Enterprise Business Group, a global unit whose job is to get large companies to put Google's AI agents into daily use. Its most concrete promise is a workforce of 1,000 forward deployed engineers, specialists who sit inside client organisations and build working applications on the Gemini Enterprise platform. Google Cloud will help train them. The unit will sit within Accenture, as part of its existing Google business group, and both companies describe it as a significant joint investment without putting a figure on it.
On its face this is a partnership announcement of a familiar kind. Its importance lies in what it reveals about the economics of artificial intelligence for the largest technology companies. Google has spent heavily on data centres, chips and models. The constraint on earning a return is no longer the technology but the slow, labour-intensive work of fitting it into corporate processes. Alphabet is now paying, in effect, to borrow the consultancy's people.
What the two companies agreed
According to the joint announcement, the new group combines Accenture's Gemini Enterprise certified staff, its forward deployed engineers, specialist Google Cloud engineering talent and Accenture's industry knowledge. It builds on Accenture's base of nearly 50,000 professionals with Google Cloud skills. Its stated priorities are to raise adoption of Gemini Enterprise with purpose-built accelerators, to build repeatable solutions for specific industries, to move clients from experiments to company-wide transformation, and to increase day-to-day use of the tools by employees.
Julie Sweet, Accenture's chair and chief executive, said the group would help clients reach AI outcomes faster by bringing together the talent, AI and data capabilities and industry expertise needed to create value at scale. Thomas Kurian, chief executive of Google Cloud, called agentic AI a top priority for enterprises and said the group significantly expands the resources available to customers. The Wall Street Journal reported the plan shortly before the companies confirmed it.
The one named result in the announcement is an internal one. YouTube used a Gemini Enterprise customer service agent during a surge in demand for NFL Sunday Ticket, and the companies say customer sentiment rose 11% while average handling time fell 37%. As TNW pointed out, YouTube is owned by Alphabet, so the flagship case study is Google deploying its own product on itself.
Why Google needs the help
Google Cloud generated $24.8bn of revenue in the second quarter, much of its growth driven by enterprise AI. Behind that sits a vast set of obligations: TechCrunch, citing reports, put Alphabet's purchase commitments and contractual obligations at $811bn at the end of June. Spending on that scale only makes sense if corporate demand for AI services keeps rising for years.
The evidence on adoption is mixed. Ramp, the corporate card company, published August data suggesting that Google accounted for roughly 6% of enterprise AI spending among its American customers, against 43.5% for Anthropic and 39.7% for OpenAI. A Google spokesperson objected that Ramp's customer base excludes many of the large enterprises signing strategic deals with Google Cloud, which is a fair caveat. Still, the figures suggest Google is not the default choice when companies buy AI tools directly. Accenture's own research, cited by ITPro, found in April that only one in ten British organisations had deployed or scaled AI in core operations.
Forward deployed engineers, a model popularised by Palantir, are the industry's answer. Rather than selling software and waiting for customers to work out how to use it, the vendor sends people to build the first applications on site. Google has already moved in this direction this year, with a $750m commitment to its partner ecosystem that placed its own engineers inside Capgemini, Cognizant and Deloitte, and a partnership with CVC Capital Partners covering the investment firm's portfolio companies. The Accenture unit is the largest step so far.
Everyone has reached the same conclusion
Google is following rather than leading. According to TNW, Microsoft launched its Frontier programme in July, committing $2.5bn and 6,000 engineers to enterprise AI deployment, two days after Amazon Web Services announced a commitment of similar scale. Anthropic and Blackstone created Ode, a $1.5bn venture that sells implementation rather than models, and OpenAI has set up a separate services business. TCS said it would turn up to 8,900 staff into deployment engineers.
Accenture, notably, works with all sides. It launched a similar Microsoft practice in March, an initiative with ServiceNow in May and a joint programme with SAP in June. For the consultancy, the Google unit adds another platform to sell. For Google, it means the most important channel to large enterprises is shared with its competitors.
There are details still missing. Neither company has said how many of the 1,000 engineers will be new hires rather than existing staff, and eWeek noted that Gemini Enterprise's agent platform supports more than 200 models, including Anthropic's Claude family and Meta's Llama. Clients building on Google's platform will not necessarily be using Google's models.
The economics behind the headcount
The deeper point concerns margins. Software earns high returns because it scales without adding people. A platform that needs teams of trained engineers inside each customer behaves more like a services business, at least during the adoption phase. TNW described a typical Accenture engagement: a team that spent eight to 12 weeks mapping a single invoice process at one client, building an agent and then handing it over to other staff to scale.
If that is the unit of work, enterprise AI revenue will arrive more slowly and at lower margins than the most optimistic forecasts assume. Hyperscalers can tolerate that, because each deployment also consumes their cloud capacity for years. The risk is to the timetable on which investors expect returns on capital spending.
Our view
The Accenture deal is a rational move and an admission at the same time. Google has some of the strongest AI models and the infrastructure to serve them, yet it is paying a consultancy to persuade companies to use them. That says the bottleneck in enterprise AI is now people and process, not technology, and it will remain so for some time.
For founders, the implication is twofold. Start-ups that sell implementation, evaluation and integration services are working in the part of the market where the giants admit they need help, and they will find buyers among both hyperscalers and consultancies. Start-ups selling thin applications on top of models face a harder fight, because the giants are now funding armies of engineers to build those applications for free as part of the platform sale. Investors should watch for named external customers and disclosed contract values in the coming quarters. Until those appear, the 1,000 engineers measure intent, not demand.
Sources
- Google Cloud Press Corner: Accenture and Google Cloud deepen partnership with formation of new Accenture Gemini Enterprise Business Group
- TechCrunch: Google Cloud races to catch up in the AI deployment wars with Accenture deal
- TNW: Google and Accenture are putting 1,000 engineers inside client offices
- ITPro: Google Cloud and Accenture launch new business group to embed forward deployed engineers with customers
- eWeek: Gemini Enterprise gets a 1,000-engineer plan from Google and Accenture
Filed under Big Tech. Written by Invest Times Editorial under our editorial standards. Commentary and analysis for information only, not investment advice.
