Everyone is talking about bigger AI models, larger GPU clusters, and the race to build the next breakthrough foundation model. But as enterprise AI matures, a different question is becoming far more important: where should AI actually run?
In the latest video on the CloudFest YouTube channel, Andrej Radonić, Founder and CEO of interSales AG, explores why the future of enterprise AI won’t be Cloud-only. Instead, it will be hybrid, distributed, and driven by workload placement rather than the unlimited fuel source known as “hype”. It’s an architectural perspective that cuts through the AI noise and focuses on what matters for Cloud providers, hosting companies, data center operators, and infrastructure specialists.
Watch the full video here:
Here are three of the biggest takeaways:
The conversation is shifting from AI access to AI placement
Public Cloud made AI accessible. It allowed organizations to experiment quickly without investing in their own infrastructure, accelerating adoption across almost every industry.
But as AI moves into everyday business operations, the questions become more complex. Organizations now have to consider latency, recurring inference costs, regulatory compliance, data sovereignty, and operational control. Rather than asking simply which model to use, businesses are increasingly asking where each AI workload should run to deliver the best combination of performance, security, compliance, and cost.
Inference—not training—will shape enterprise AI economics
While model training attracts the headlines, Andrej argues that inference is where businesses will spend most of their money. Every AI-powered search, customer interaction, coding assistant, or document summary becomes part of an ongoing operational cost rather than a one-time investment.
That makes infrastructure decisions increasingly important. Choosing the right environment for each workload can improve performance, reduce costs, strengthen governance, and deliver a better user experience. For infrastructure providers, this represents a major opportunity to help customers optimize AI beyond simply providing compute.
Hybrid AI creates new opportunities for infrastructure providers
The future isn’t public Cloud versus private infrastructure. It’s about using the right environment for the right workload.
Some AI applications will always belong in hyperscale Cloud environments, while others will benefit from running closer to users, closer to data, or inside highly-regulated private infrastructure. For Cloud providers, hosting companies, and data center operators, this shift creates opportunities to deliver differentiated services around sovereignty, compliance, latency, and specialized AI infrastructure rather than competing directly with hyperscalers.
Stay ahead of what’s next
CloudFest is where the internet infrastructure community explores the technologies shaping tomorrow’s digital economy.
Subscribe to the CloudFest YouTube channel for more expert analysis, interviews, and technical insights.
PLUS… sign up for the CloudFest newsletter to stay up to date with the latest thinking on AI, Cloud, hosting, connectivity, and the infrastructure powering it all.
