Google Cloud bets big on AI agents with Gemini Enterprise platform and new chips

Google Cloud pushes AI agents as core enterprise strategy

Google Cloud is placing AI agents at the center of its enterprise ambitions, signaling a strategic shift as pressure mounts to demonstrate returns on massive AI investments. At the Google Cloud Next conference in Las Vegas, CEO Thomas Kurian and Alphabet CEO Sundar Pichai positioned the company's AI tools as production-ready infrastructure for enterprise customers, who represent the industry's most reliable revenue stream 4

4

. The tech giant announced it's chasing the vision of the "agentic enterprise," with Google Cloud reporting that 75% of its customers already use AI in their businesses 1

1

. Agentic AI powers autonomous bots that can complete tasks with minimal human oversight, transforming workflows particularly for coding and workplace automation.

Gemini Enterprise Agent Platform unifies Google's AI offerings

The centerpiece of Google's announcements is the Gemini Enterprise Agent Platform, which consolidates what was previously known as Vertex AI along with new security, governance, and orchestration features 5

5

. This behind-the-scenes technology allows businesses to oversee all their AI agents from a single platform. Employees can create and deploy agents through the Gemini Enterprise app without writing code, using a new agent designer that schedules tasks across different applications 1

1

. The platform includes Memory Bank and Memory Profile features to help agents remember past interactions with users, addressing a weakness in early AI tools 2

2

. Agent Simulation enables developers to test tools thoroughly before launch, while Projects serves as a collaboration platform bringing together information from Workspace, Microsoft's OneDrive, and company chats 2

2

.

Eighth-generation TPUs target training and inference workloads

Google unveiled two new eighth-generation TPUs—the 8T and 8I—advanced chips designed exclusively for heavy-compute tasks like developing AI 1

1

. The 8T chip delivers three times the processing power of Google's seventh-generation Ironwood for AI model training, while the 8I specializes in inference with an 80% improvement in SRAM memory and approximately 11,152 chips in a single system 1

1

. These Tensor Processing Units are central to Google's full-stack AI approach, which Kurian describes as a competitive advantage: "We're not just a hyperscaler reselling other people's technology. Our differentiation comes down to the fact that we own the IP, the model and the chips are ours" 3

3

. This vertical integration means Google retains more revenue instead of paying chip or model providers, allowing greater reinvestment.

Closing the gap in cloud computing market share

After a slow start, Google Cloud has grown its cloud market share from 7% to 14% over eight years, though it remains a distant third to Amazon Web Services and Microsoft Azure in the $418 billion cloud computing market 3

3

. AI is accelerating growth—Google Cloud reported a 48% revenue jump in the final quarter of 2025 and is on track to generate more than $70 billion this year, up from $43 billion in 2024 3

3

. Kurian credits the company's 12-year investment in DeepMind for allowing continuous improvement of proprietary chips and delivery of consumer and enterprise AI products at lower cost with better margins 3

3

. Epoch AI estimates Google controls about a quarter of global AI computing power, roughly 3.8 million TPUs and 1.3 million GPUs, ahead of Microsoft's 3.2 million Nvidia GPUs 3

3

.

Strategic positioning to compete with OpenAI and Anthropic

Google faces intense competition as it works to justify pouring as much as $185 billion into capital expenditure this year alone 2

2

. A particular focus is AI coding, where company leaders worry they've fallen behind—many Silicon Valley engineers toggle between Anthropic's Claude Code and OpenAI's Codex, but Google often isn't part of that conversation, startup founders told Bloomberg News 2

2

. While rivals push coding products aggressively, Kurian kept coding largely out of the spotlight, instead casting the battleground as one defined by agents, governance, and enterprise deployment 4

4

. Google is pitching Gemini Enterprise as more ready for large business needs than Anthropic's Cowork and other agentic software 5

5

. The company also introduced cybersecurity agents to help clients protect systems as AI models identify torrents of bugs 2

2

.

Long-term viability and market shakeout ahead

Kurian argues Google's integrated approach positions it favorably as the market matures, particularly as OpenAI and Anthropic face financial pressures. Both startups are losing tens of billions annually while racing to secure computing power, and Kurian suggests private capital markets are reaching saturation: "If you're going to go public, you can't be lossmaking forever. And if you stay private, you cannot raise venture money forever" 3

3

. This year, OpenAI and Anthropic raised more than $150 billion in two of the largest private fundraisings in history as they prepared for IPOs 3

3

. Kurian predicts market consolidation: "Over the next year to two you will see some shakeout in the market. Whether particular providers are going to make it or not largely comes down to the economics" 3

3

. For enterprise customers, Google's comprehensive backbone combining chips, foundation models, and developer tools represents an alternative to cobbling together individual services, with the promise of better security, compliance, and cost optimization as businesses automate business tasks at scale 2

2

.