Google Bets $750M on Consultants Building Your AI Agents, and a Robot Just Beat Elite Athletes at Their Own Sport

Google Cloud announced a $750 million innovation fund at Cloud Next 2026 to accelerate agentic AI development across its partner ecosystem. The fund targets consulting firms and systems integrators: Accenture, Deloitte, KPMG, PwC, Cognizant, HCLTech, TCS, and NTT DATA. The investment covers AI value assessments, Gemini proofs-of-concept, agentic AI prototyping, forward-deployed engineering teams embedded alongside consulting firms, and Wiz security assessments. Google DeepMind will give early access to Gemini models to select partners, who will use the tools and provide feedback ahead of launch. Bloomberg reported that McKinsey is also among the beneficiaries. (Google Cloud Blog · Bloomberg · Channel Dive)

The partner commitments in return are substantial. Accenture has built more than 450 agents on Google Cloud and is expanding its Gemini practice across all industry verticals. Deloitte described its investment as the "largest yet" in any single cloud AI platform and has deployed more than 100 agents for enterprise customers. KPMG committed $100 million of its own capital to build agentic AI solutions on Google Cloud. PwC announced a $400 million collaboration focused on security and compliance agents. NTT DATA has dedicated 5,000 engineers to Google Cloud agent development. Thomas Kurian, Google Cloud CEO, told Reuters: "There's definitely a strategic shift as the models become much more sophisticated." Alongside the fund, Google announced Gemini Enterprise Agent Platform, its 8th-generation TPUs (TPU 8t for training, TPU 8i for inference), and an Agentic Data Cloud with a cross-cloud AI-native lakehouse. (The Next Web · Reuters via Taipei Times · Google Cloud Press)

This is Google's clearest signal that the agentic AI business model is not API subscriptions sold to developers; it is consulting engagements sold to enterprises. The $750 million fund is a subsidy to make Google Cloud the default platform for the agents that consulting firms build for their Fortune 500 clients. If Accenture builds 450 agents on Google Cloud and Deloitte calls it their largest cloud AI investment, the lock-in operates at the consulting layer, not the model layer. That is a strategically distinct bet from what OpenAI and Anthropic are pursuing. OpenAI sells model access. Anthropic sells model access plus safety. Google is selling the entire deployment stack through the firms that already have enterprise purchasing relationships.

We covered Cloudflare's Project Think on April 21, which targets the other end of the market: cheap, fast, secure agent execution for developers and consumer products. Google's $750 million fund targets the enterprise end. The agentic infrastructure market is bifurcating along the same line: developer-first (Cloudflare, Anthropic) vs. enterprise-first (Google Cloud, Microsoft).

Investment signal: The consulting firms' commitments (Accenture at 450+ agents, KPMG at $100M, PwC at $400M) are the leading indicator. When the Big Four commit nine-figure sums to a single cloud platform's agent stack, the enterprise agentic AI market is no longer speculative. The question is which platform captures the consulting layer.

Roadmap signal: If your company buys consulting services from any of these firms, expect agentic AI proposals built on Google Cloud to arrive in the next quarter. The fund's forward-deployed engineering teams mean Google engineers will be in the room during architecture decisions.