The age of artificial intelligence agents has arrived. From the neon-lit corridors of a Las Vegas cloud conference to the factory floors of Osaka, two converging stories this week underscore that autonomous AI systems are transitioning from corporate pilot programs into the foundational layer of global enterprise.
Key takeaways
- Google has officially ended the experimental phase of enterprise AI, consolidating its agent strategy under “Gemini Enterprise” and backing it with $175–185 billion in capital spending and new TPU 8t/8i chips purpose-built for agentic workloads.
- Japan’s AI agent market is set to grow nearly tenfold from $250 million in 2024 to $2.43 billion by 2030, driven less by hype than by demographic necessity as the country confronts a shrinking workforce and an aging population.
- Across both markets, agentic AI has crossed the threshold from pilot to deployment, with Forrester documenting 55–75% process cycle-time reductions and the global enterprise agent market projected to reach $142.35 billion by 2035.
The question that once dominated boardroom discussions, whether agentic AI was ready, has quietly been retired.
Google Draws a Line in the Sand
At Alphabet’s annual Google Cloud Next conference in Las Vegas, CEO Sundar Pichai and Google Cloud chief Thomas Kurian made no effort to temper expectations. Kurian set the tone during the opening keynote, declaring that the experimental phase is now behind the industry and that the real challenge is just beginning. The message was deliberate: for enterprise customers, the moment for cautious evaluation has passed.
Google signaled to investors that AI agents, human-like digital assistants capable of autonomous planning, decision-making, and action, are the linchpin of its strategy to monetize artificial intelligence, with large enterprise customers emerging as the industry’s most reliable revenue stream.
The financial commitment backing this shift is substantial. Pichai reaffirmed Alphabet’s capital spending plan of between $175 billion and $185 billion for the year, with just over half of the company’s investment in machine learning computing power dedicated to its cloud business.
To consolidate its position, Google announced it was unifying a suite of AI products under the name “Gemini Enterprise,” most notably rebranding and expanding Vertex AI, a platform that allows cloud customers to select from a range of AI models for business applications. According to Kurian, the platform’s primary use case has already undergone a quiet but dramatic transformation, shifting away from traditional machine learning workflows toward a surge in users building custom AI agents.
The hardware strategy is equally ambitious. Google unveiled two new custom tensor processing units, the TPU 8t and TPU 8i, both designed end-to-end for what the company describes as the age of agents. The TPU 8i, tuned for inference workloads that power real-time agent responses, delivers 80 per cent better performance than the prior generation.
Google’s push is also a direct response to intensifying competition. While traditional cloud rivals such as Amazon and Microsoft remain ahead in overall market share, Google has grown its cloud market share to 14 per cent as of end-2025, buoyed by heavy investment in AI, data centres, custom chips, and networking infrastructure. Meanwhile, model providers including OpenAI and Anthropic have aggressively shifted resources toward enterprise customers, pushing into application-layer tools and agent-building platforms, forcing Google to differentiate not on model capability alone, but on end-to-end enterprise infrastructure.
Japan: Where Demographics Are Driving a $2.43 Billion Imperative
While Silicon Valley frames agentic AI as a strategic opportunity, Japan’s adoption is being shaped by something far more urgent: structural necessity.
Japan’s AI agent market was valued at $250 million in 2024 and is projected to reach approximately $2.43 billion by 2030, growing at a compound annual growth rate of 46.3%, the highest among all enterprise software verticals in the country, and one of the fastest in the Asia-Pacific region.
The driver is not technological enthusiasm. It is demography. With more than 28 per cent of Japan’s population aged 65 or above and the working-age population shrinking for over two decades, the country faces a fundamental gap in manpower that immigration reform and productivity initiatives cannot adequately bridge. AI agents, autonomous software systems capable of continuous operation without human supervision, offer the most viable solution at scale.
Government policy has hardened this trajectory into law. Japan adopted the Act on the Promotion of Research and Development and Utilization of AI-related Technologies in May 2025, establishing an AI Strategic Office at the Cabinet Office under direct oversight of the Prime Minister and underpinning a ¥3 trillion public-private partnership for frontier AI infrastructure. The country’s broader Society 5.0 framework, a government blueprint for a human-centric, technology-integrated society, has positioned AI agents as instruments of national industrial policy, not just commercial innovation.
Sector-level adoption reflects Japan’s industrial heritage. Manufacturing and industrial automation represents the largest current segment, with AI agents deployed for predictive maintenance, quality control anomaly detection, and supply chain coordination, functions that align naturally with Japan’s precision-manufacturing culture. Healthcare is the fastest-growing vertical, driven by the mounting pressures of an aging population, with agents already deployed in hospital networks for patient triage, medication management, and clinical documentation.
The competitive landscape is increasingly crowded. Domestic heavyweights including Fujitsu, NEC, Hitachi, and NTT Data have each launched enterprise AI agent platforms in the past year, while SoftBank, NEC, Sony, and Honda have partnered to develop a sovereign Japanese AI model with a capacity of one trillion parameters, a direct challenge to US and Chinese frontier AI dominance. Global players are moving in parallel: Microsoft pledged $2.9 billion in Japanese AI and cloud investment in April 2024, while NTT Corp. released its Smart AI Agent Ecosystem backed by $59 billion in five-year domestic commitments.
Convergence: A Global Infrastructure Shift
What emerges from these two narratives, Google’s enterprise pivot and Japan’s structural adoption, is a coherent picture of an industry crossing a threshold. The terminology has changed. Vendors no longer speak of AI “pilots” or “proofs of concept.” The word now is deployment.
Google’s Kurian told Reuters that the shift reflects models becoming significantly more sophisticated, with enterprises now building custom agents rather than merely experimenting with off-the-shelf tools. In Japan, research firm Forrester has found that enterprises deploying agentic AI workflows are achieving process cycle time reductions of between 55 and 75 per cent in relevant use cases, numbers that make the business case increasingly difficult to ignore.
The broader global enterprise AI agent market, valued at $6.65 billion in 2025, is forecast to reach $142.35 billion by 2035, with Japan’s trajectory outpacing even that remarkable global average. As Google bets its infrastructure roadmap on agents and Japan bets its economic future on them, the remaining question for enterprises worldwide is no longer whether to act, but at what speed, and with whose tools.
