Thomas Kurian, CEO of Google Cloud, spoke at the Goldman Sachs Communacopia + Technology Conference on September 9, 2025, highlighting how AI is reshaping enterprise cloud adoption and positioning Google Cloud for long-term growth. He noted that while cloud penetration is still in its early phases, enterprise priorities have shifted from application hosting to business transformation powered by AI.
Kurian emphasized Google’s strategy of building end-to-end differentiation—from custom silicon and AI-optimized networking to its Gemini model suite and enterprise-focused agents. He said Google Cloud now supports nine of the ten largest AI labs globally and continues to see momentum across enterprises, financial services, and high-performance computing users. The company’s $106 billion backlog, with more than half converting to revenue in two years, reflects strong customer commitments.
Kurian also outlined a diversified monetization model for AI and showcased examples of enterprises deploying Gemini, diffusion models, and domain-specific agents. “We’re not just reselling third-party technology. We’ve built chips, models, and software of our own, and customers are adopting it at scale,” he told the audience.
• Cloud adoption remains early: most workloads still on-premise, with sector differences in migration pace (govt. slower; Europe constrained by sovereign cloud rules).
• Differentiation pillars: performance, cost, reliability, and efficiency in AI infrastructure.
• Google AI systems offer ~50% higher performance vs. competitors, 2x FLOPS per watt, and 118x more throughput per system.
• AI-optimized storage driving 37x increase in data volumes; optical networking allows dynamic reconfiguration of training vs. inference clusters.
• Software stack leadership includes JAX, XLA, Pathways; delivers 33x inference efficiency gains in past year.
• Customer demand spans four groups:
– AI labs (9/10 largest are customers)
– Enterprises deploying AI (e.g. LG, ServiceNow, Canva)
– Capital markets moving to AI-based inference
– HPC/scientific research (e.g. SSI, LG AI)
• Gemini models: used by 9M developers; 2.5 reached 1T tokens 20x faster than Gemini 1.5.
• Model portfolio includes diffusion (images, video, audio), time series, and molecular design.
• Agent Development Kit supported by 120+ companies; Agentspace platform integrates domain-specific agents.
• Agents adoption: Gemini CLI agent at 1M users since June; customer service interactions up 10x; commerce agent processes 5B transactions.
• Data Cloud growth: BigQuery usage with Gemini up 27x; handling structured + unstructured data; used by Radisson Hotels, Virgin Media.
• Security: protecting data, models, and against AI-driven threats; 1.5B+ Gemini-powered “threat hunts.”
• Monetization: consumption (GPU/TPU/models by token), subscriptions (Workspace, Agentspace), usage growth, value-based pricing (e.g., ad conversions, call-center deflections), and upselling.
• Revenue diversification: many product lines, all growing; 65% of customers using AI tools use 1.5x more products overall.
• Backlog at $106B, growing faster than revenue; >50% to convert to revenue in 2 years.
• Operating discipline: capital efficiency through scale, fleet optimization, water cooling, and lowest global PUE; improving go-to-market efficiency by focusing on existing customers.
• Partnerships: close collaboration with Nvidia; open stack approach with support for third-party accelerators and models.
• Investment priorities: capital infrastructure (data centers, power, sovereignty compliance), product domains, and go-to-market specialization.
• AI adoption patterns: four domains—digital products (Snap, Warner Bros.), customer service (Verizon, Wendy’s, Mercedes), back-office (Home Depot, AES, Tyson), and IT (code generation, cybersecurity).
• Six-month model revision cycles fueling new products, e.g., Veo 3 video creation driving demand from media and advertising sectors.
🌐 Analysis: Google Cloud is leveraging vertical integration—custom silicon, optical networking, and native models—while offering openness with Nvidia and third-party AI models. Its strong share in AI labs and expanding enterprise adoption contrasts with AWS’s infrastructure-first model and Microsoft’s deep SaaS integration. With $106B in backlog and rapid uptake of Gemini, Google Cloud is betting its dual strategy of infrastructure efficiency and domain-specific agents will drive outsized growth into 2026.
