Google Cloud and Accenture formed a joint unit to push Gemini into enterprises
Google will train up to 1,000 Accenture consultants as forward-deployed engineers. It is trying to close a gap that Ramp data puts at 6% of US enterprise AI spending against Anthropic's 43.5%.
Google Cloud and Accenture have formed the Accenture Gemini Enterprise Business Group, a joint unit that will place forward-deployed engineers inside enterprises to build on Google's AI platform. Google will train up to 1,000 Accenture consultants for the role, and the unit sits within Accenture's organisation.
Terms were not disclosed. It follows a $750 million partner ecosystem commitment Google Cloud announced earlier in the year.
The number behind the urgency
Google Cloud is not short of revenue — $24.8 billion in the second quarter, substantially driven by enterprise AI. It is short of share in the specific market for model consumption.
Ramp data from August cited by TechCrunch puts Google at roughly 6 percent of US enterprise AI spending, against Anthropic at 43.5 percent and OpenAI at 39.7 percent. Google notes, reasonably, that the Ramp figures exclude large strategic deals with the likes of Oracle, Meta and ServiceNow, so the real gap is narrower than 6 percent implies.
It is still a gap, and it is not a capability gap. Google shipped Gemini 3.8 Flash on 2 September, its third Flash model in six weeks, alongside a security variant claiming more than 70 percent on internal vulnerability discovery across 20 languages and 2.6 times more correct patches to Chrome vulnerabilities than leading commercial models.
The models are competitive. The deployment is not.
Forward-deployed engineering is now the whole competition
The pattern is unmistakable this year. Microsoft launched an FDE practice with Accenture in March. OpenAI operates a unit it calls The Deployment Co. Amazon and Anthropic have both launched dedicated FDE teams. Kirkland & Ellis went to Palantir — the company that invented the model — rather than to a legal AI vendor.
The reason is that enterprises have run out of road on self-service. A company can buy API access in an afternoon and still be unable to connect a model to its own systems, because the work is data plumbing, access control, evaluation and workflow redesign, not prompting.
Caterpillar's $100 million programme to train 118,000 employees over five years rests on the same insight: its own framing is that deployment depends on experienced operators redesigning workflows rather than on installing models.
So the constraint on AI revenue is now the supply of people who can do integration work, and every major vendor is buying that supply. Accenture is the largest pool of it, which is why Microsoft and Google have both now contracted with the same firm.
The problem with renting your channel
The unit sits inside Accenture. Google is training consultants who work for a company that also has an FDE practice with Microsoft, and whose commercial interest is in whichever platform the client will pay to implement.
An integrator is not a captive channel. It is a mercenary one, and the platform that wins a given engagement will often be the one the client already uses for email.
Google's counterweight is cost and vertical integration — it owns TPUs, and can price aggressively. Its liability is trust: Alphabet carries $811 billion in infrastructure purchase commitments and just avoided a forced sale of AdX, with Judge Leonie Brinkema instead ordering behavioural interoperability remedies. Enterprise buyers weighing a strategic dependency on Google notice both.
Neither company disclosed revenue targets, the duration of the arrangement, or whether the 1,000 trained consultants are exclusive to Gemini work.
Runs the newsroom. Rename this profile in the studio to your own byline.
Related
Every weekday, the AI stories that moved money or shipped code.
No cross-posting, unsubscribe anytime. See all newsletters