With global cloud spending surging past $102.6 billion in a single quarter, it’s clear that enterprise AI has moved from the laboratory to the core of business strategy. This monumental 25% year-over-year growth is being driven by companies transitioning from isolated experiments to full-scale AI deployments. To help us understand this pivotal shift, we are speaking with Dominic Jainy, a leading IT professional and analyst who specializes in the practical application of artificial intelligence and machine learning within the enterprise. We will explore the challenges of scaling AI, the evolving battleground among cloud hyperscalers, and what the massive financial commitments from enterprises signal for the future of the market.
Global cloud spending hit $102.6bn in Q3 2025, reportedly driven by enterprises moving from AI experimentation to scaled deployment. Could you walk us through the practical steps and key challenges companies face when making this transition from a simple AI proof-of-concept to a full-scale, integrated deployment?
It’s a journey that looks simple on paper but is incredibly complex in reality. The initial proof-of-concept is often an exciting, isolated success. But scaling it is where the real work begins. The biggest hurdle we’re seeing is that many enterprises simply lack the standardized building blocks needed for a production environment. When you move from a handful of users to thousands, you suddenly have to worry about business continuity, ensuring the AI agent doesn’t fail during peak hours. You have to consider the customer experience, making sure it’s seamless and reliable. And most critically, you have to manage compliance and governance at scale. This gap in foundational, standardized components is what’s genuinely slowing the real-world deployment of AI agents, turning what should be a straightforward path into a significant engineering and organizational challenge.
The report mentions competition is shifting from model performance to platform capabilities for AI agents. What specific platform features, beyond simple performance metrics, are essential for enterprises to reliably run and manage multi-model AI agents, like those mentioned in connection with Amazon Bedrock or Google’s Vertex AI?
This is the most important evolution in the market right now. For a long time, the race was about who had the “smartest” model. Now, that’s just table stakes. Enterprises have realized that being locked into a single model is a huge risk. What they desperately need are platform capabilities that provide flexibility and control. The key feature is robust multi-model support, which is now seen as a production requirement, not just a nice-to-have. This allows a business to switch between models for resilience if one provider has an outage, for cost control by choosing the most efficient model for a given task, and for deployment flexibility to use the best tool for the job. Beyond that, platforms like Bedrock and Vertex AI are winning by offering enhanced safety measures, data automation tools, and what the industry is calling “agent build-and-run capabilities” to manage the immense complexity of deploying these systems securely and reliably in the real world.
The big three hyperscalers now account for 66% of the market, yet Azure and Google Cloud grew significantly faster than AWS in Q3. What specific strategies, particularly around enterprise AI offerings like Gemini Enterprise or the OpenAI partnership, are fueling this rapid growth for Microsoft and Google?
While AWS is still the market leader with a 32% share, the incredible growth rates of Azure at 40% and Google Cloud at 36% are directly tied to their aggressive and well-executed enterprise AI strategies. For Microsoft, the deep, renewed partnership with OpenAI gives them immense credibility and a powerful narrative. Enterprises see Azure as a direct conduit to cutting-edge AI. Google, on the other hand, is leveraging its own powerful research and building a comprehensive, integrated platform. The launch of Gemini Enterprise is a perfect example. They aren’t just selling access to a model; they’re providing a full-stack solution with enterprise-grade agents, no-code development tools, and critical security and governance built-in. This all-in-one approach is resonating strongly with businesses that want a clear, supported path to deploying AI, which is fueling that 36% growth.
The report notes rising backlogs for providers like AWS ($200bn) and Google Cloud ($157.7bn) signal market resilience. Could you break down what these backlogs represent in terms of customer commitments and what specific, high-value services enterprises are locking in long-term to create such massive figures?
Those staggering backlog numbers—$200 billion for AWS and a rapidly growing $157.7 billion for Google—are a powerful indicator of where the market is headed. These aren’t just short-term purchases; they represent multi-year, strategic commitments from large enterprises. They are essentially pre-paying for capacity and services, locking in their digital transformation roadmaps with a specific cloud provider. While this certainly includes foundational infrastructure like compute and storage, a significant and growing portion of these commitments is for high-value AI and data services. Companies are making long-term bets on platforms like Amazon Bedrock or Google’s Vertex AI. They are committing billions to ensure they have the tools, models, and platform capabilities they will need over the next several years to build and deploy their generative AI workloads, which shows a tremendous level of confidence and strategic alignment with their chosen cloud partner.
What is your forecast for the evolution of enterprise AI on cloud platforms over the next 18-24 months?
Over the next two years, the battle will be won and lost on the strength of AI platforms, not just models. The focus will intensify on making the deployment of sophisticated, agentic AI not just possible, but practical and manageable for mainstream enterprises. I forecast that the “agent build-and-run” capabilities will become the primary competitive differentiator. We will see hyperscalers racing to offer the most comprehensive, low-code, and secure environments for building, testing, and governing multi-model AI agents. The concept of using a single AI model for all tasks will seem archaic; multi-model support will be the absolute baseline. Ultimately, the provider that can best abstract away the underlying complexity and empower businesses to solve real-world problems with AI agents, safely and at scale, will capture the next wave of growth.