Access is the hook that ties Uli Homann’s professional focus together. It’s his belief that when everyone has access to AI capability and systems can access each other, innovation compounds. In the era of agentic AI, therefore, openness is the critical foundation for scale.
Corporate Vice President and Distinguished Architect - Microsoft
Uli sees the last three decades of computing as evidence for the idea of ever-expanding access. Client-server made enterprise software affordable and reachable beyond mainframes. Cloud hosted software removed hosting complexity so more people and organizations could use richer services and solutions. And now, AI is scaling cognition to democratize knowledge in an unprecedented way.
He frames the evolution of AI in three phases – each expanding both access and tangible business ROI.
We’ve already seen GenAI transform productivity tools, which then led us to the team member phase – where people come to rely on AI tools like CoPilot as peers in business. And now we are entering the autonomous phase, where agents drive processes, with humans as supervisors.
Now, agentic AI can monitor and manage complex scenarios, such as supply chain disruptions, by dynamically formulating and executing plans based on real-time data and knowledge without pre-programmed responses.
Crucially, this isn’t yesterday’s automation. Traditional systems only handle what a developer coded upfront, but agentic AI uses knowledge and live data to form and execute plans towards a declared goal – with permission and capability to adapt, not just read scripts.
This is what happens when individuals and enterprises are given access to tools that enhance productivity, capability, and innovation.
Access at this scale is not without its challenges. And Uli is clear about the biggest brakes on AI success – three critical barriers crashing together, each challenge compounding the next: energy, GPU, geography.
First, energy constraints are the primary bottleneck for AI at scale, data center capacity is currently insufficient for projected AI workloads, and existing methods of securing energy are unsustainable. As a result, Microsoft is investing in advanced liquid cooling, water-sparing designs and next-gen nuclear partnerships to secure low-carbon power at scale. Critically, the goal isn’t more watts – it’s responsible watts.
Second, with market demand and supply chain limitations complicating deployment planning and scalability, planning GPU availability becomes a puzzle of budgets, timelines and locality rules.
And finally, the next phase of AI requires input from everyone, not just a select few. But much of the world’s AI capability is clustered in a few regions, which is something he’s keen to change. “We’re building AI data centers globally, so countries aren’t locked out of the next wave of productivity,” he says.
It’s this access imperative that Uli’s work is particularly grounded in – and for Uli, transparency is the key in making innovations in AI accessible to everyone.
With agentic systems gaining the capacity to act, Uli believes this new level of access to data must be governed with transparency, guardrails, and responsibility.
Uli’s current focus is on building a framework to help enterprises manage AI agents safely and consistently across vendors and platforms. He believes that “without shared standards, there’s no way you can scale.”
He’s also leading Microsoft’s contributions in OpenTelemetry, an open-source observability standard, by adding an agent scope – a common way for AI agents to share what they’re doing and why.
“If you can’t see the behavior of your agents, you can’t govern them. Open observability is the backbone of trust.”
This alone is a technical breakthrough with far-reaching implications for industries reliant on large-scale infrastructure. In the same way that network visibility transformed cybersecurity in data centers, and how open experimentation gave us the internet, Uli believes agent observability will define the next generation of enterprise AI operations.
For Uli, this is simply another turning point in how technology grows – not through competition, but through collaboration.
“We are seeing the same wave in agentic AI. Without interoperable standards where every agent behaves in a certain predefined way. As a result, businesses can adopt agentic AI, but not at scale.”
That same spirit of collaboration is what Uli believes will drive true progress in the data and telecom ecosystem – from AI-driven workload management in data centers to cross-industry governance standards that ensure safety and accessibility.
“Every time we connect systems, we connect minds, that’s where the real innovation begins.”
Uli’s vision for what’s next is clear. Openness and access are no longer soft ideals – they’re infrastructure on which intelligence scales. And it’s this advice he gives to young engineers entering the workforce today.
“The fundamentals of computer science are still critical. But we need people who are curious and willing to engage in experimentation every day. Be less critical and be more open.”
Whether in a hyperscale data center or a single autonomous system, the principle remains the same: the more connected we are, the more capable we become.
From the PC era, the rise of Cloud and now the AI boom – few but Uli can say they’ve been at the helm of almost all technological revolutions of the digital age. It’s a career forged across an unlikely path: from police officer in Germany to a 35-year career at Microsoft. The moment that changed everything? Learning to program.
“You express a thought in code, the compiler accepts it, and the machine does what you intended. That’s just pure magic.”
He’s been chasing that feeling and turning ideas into shared capability ever since.
- Articles
Do you want to feature in the next edition of Uniquely Wired? Our editors are always looking for new stories. Is yours exciting?