Agenda
Tuesday, 8 September 2026 — Welcome & Connect | 18:30 - 20:30
Join us on the Hotel Arts terrace for our opening welcome reception. Connecting with fellow AI infrastructure leaders from around the world while enjoying a view of the Mediterranean. Designed to foster meaningful conversations from the outset, the evening provides an opportunity to meet peers, partners, and industry leaders before the formal program begins.
- Executive Welcome Reception
- Networking with executive peers, NVIDIA, and AI ecosystem partners
- Informal discussions to kick off the summit
Wednesday, 9 September 2026 — Building a Profitable AI Business | 08:00 - 21:00
Grab a coffee, get registered, and start making connections before the day kicks off.
Abstract
Another chance to connect. Pre-booked meetings happen now, and everyone else is welcome to keep mingling and building on today's conversations.
Neysa's CPO, Karan Kirpalani, on building a business that beat the market on margins when few believed it could, and the discipline that turned into real scale. A candid read on what superior unit economics look like and how they compound.
Abstract
Abstract
Capital is flooding into AI infrastructure, but raw capacity is not a business. Haseeb sets the core idea of the summit and folds in the four directions Rafay is building toward and why. The session frames where margin lives and how operators move from selling GPU time to selling outcomes.
Under Haseeb's leadership, Rafay has become a strategic technology partner to organizations building and operating large-scale AI infrastructure, helping them accelerate time to market, improve developer access to compute, and create new revenue opportunities from GPU and AI investments. Haseeb is a frequent industry speaker on AI infrastructure, sovereign AI, AI Factories, and the rapidly evolving economics of AI services.

Abstract
Warren Barkley, VP, Product, DSX at NVIDIA, opens Rafay's customer event with NVIDIA's strategy for sovereign and neocloud AI infrastructure: the thinking behind what NVIDIA expects from the ecosystem, and what that means for operators and enterprises buying into this stack today. He'll be joined by Nicholas Chia, CEO of Aolani — one of the first operators running in production on DSX — for a conversation on what it actually takes to go from silicon to a service customers will pay for.

Abstract
Ashin Uday from Yotta joins Haseeb and Warren for a conversation on what it actually takes to go from silicon to a service customers will pay for.

Abstract
AI demand is growing rapidly, and building a sustainable neocloud business depends on both GPU access and the economics of operating them. This session connects the economics of AI infrastructure with the capabilities enterprises expect from their providers.We’ll examine the margin math behind the neocloud model—including GPU utilization, pricing, and the balance between training and inference workloads—alongside what enterprises are actually asking for from AI infrastructure operators. Drawing on advisory work across sovereign and neocloud programs worldwide, David Wood, Senior Managing Director, Global Lead, Sovereign AI, will identify where operators most often fall short and what it takes to address those shortcomings.
Connect with peers, continue conversations, make new connections, or meet with attendees you’ve pre-booked time with.
Enjoy a great lunch while connecting with peers and industry leaders. Compare notes on the morning’s key takeaways, continue the conversations sparked by the sessions, and recharge for an engaging afternoon ahead.
12:45 - 13:15 | Session: Lessons Learned from Deploying 35+ Multi-Tenant GPU Clouds and AI Factories
Abstract
Drawing on lessons from 45+ production GPU cloud and AI factory deployments, Netris CEO Alex Saroyan explores what it takes to operate complex AI networks at scale. He’ll cover the challenges of managing multiple networking layers, enabling hardware-enforced multi-tenancy, and moving beyond DIY automation to build more scalable, secure, and repeatable AI infrastructure.

A financial panel on the economics of the AI business: GPU-as-a-service margins versus token-based AI services, and how utilization, pricing, and mix drive ROI on a major infrastructure bet.
Finance and operations leaders from leading neoclouds compare where the returns are coming from today and how they are planning to grow margins from here.
Another chance to connect. Pre-booked meetings happen now, and everyone else is welcome to keep mingling and building on today's conversations.
Abstract
AI leaders are no longer asking how many tokens they can process, but what outcomes they can create with the right balance of cost, control, and trust. As agentic AI drives dramatically larger workloads, enterprises will need hybrid infrastructure spanning frontier cloud, private AI, hosted environments, and the edge. This session explores why hybrid AI is becoming the default enterprise model, and why intelligent control planes that route workloads across environments may become the winning layer.
With more than a decade at Lenovo, Noam has held leadership roles across high-performance computing, enterprise AI, cloud AI, and enterprise infrastructure. He also oversees Lenovo's European Presales Center of Excellence and European Innovation Center, supporting customers and partners with technical expertise, solution development, and AI demonstrations. Noam is a qualified attorney and holds an MBA.

Abstract
The next wave of AI will be about turning capacity into AI factories that can serve enterprises, governments and entire regions at scale.
Drawing on NVIDIA's work across supercomputing, cloud partners and the broader AI ecosystem, Rod Evans, Vice President, AI Cloud Infrastructure, EMEA, explores how the AI infrastructure market is evolving from centralized GPU capacity toward distributed, sovereign and service-driven AI factories. He'll examine what separates infrastructure that simply delivers compute from platforms capable of attracting workloads, improving utilization and creating durable economic value.
Rod’s 30+ years of international experience enables him to lead teams with passion, integrity, and confidence. He is a committed innovator with a keen interest in increasing awareness of AI’s potential. He works tirelessly to encourage the early adoption of AI by showcasing what it can do, how it works, and why implementing an AI strategy is critical.
He joined NVIDIA in 2018 and has overseen strong growth in the Artificial Intelligence businesses at NVIDIA across Europe. Before joining NVIDIA, Rod led the European direct sales force at Intel Corporation and previously held European Vice President Positions at Silicon Graphics, Rackable Systems, Copan Systems and EMC. Currently based in Reading, United Kingdom; he has worked in the UK, USA, and Germany.
Rod is a British national and alumnus of Queen Mary University, London.

Abstract
Sovereign AI as a business and a mission at once. Cassava's leadership on building toward a continent-scale AI business across Egypt, Nigeria, and Kenya. The story offers a model for turning regional reach into a defensible business rather than raw capacity.
Abstract
A fleet can be sold out, fully allocated, and still miss its margin. "Utilization" is three numbers, not one: what you sold, what got assigned to a job, and what the GPUs spent computing rather than waiting for data. Sales owns the first. Orchestration owns the second. The third has no owner and is capped at design time, months before the first customer. Izhar Sharon separates the three, names the workloads where a fast data tier adds margin and the ones where it takes margin away, and closes with five questions worth asking your own team before the next site is committed

Under Haseeb's leadership, Rafay has become a strategic technology partner to organizations building and operating large-scale AI infrastructure, helping them accelerate time to market, improve developer access to compute, and create new revenue opportunities from GPU and AI investments. Haseeb is a frequent industry speaker on AI infrastructure, sovereign AI, AI Factories, and the rapidly evolving economics of AI services.

Connect with peers, continue conversations, make new connections, or meet with attendees you’ve pre-booked time with.
Abstract
Haseeb closes on the shared playbook behind building AI as a business and, for many, an act of sovereignty.
Enjoy dinner with your peers, continue the day’s conversations, and make new connections in a relaxed setting.
Thursday, 10 September 2026 — Leading the Next Generation of AI Businesses | 08:30 - 16:00
Fuel up and mingle. Connect with familiar and new faces before things kick off.
Abstract
The people, the mission, the vision. Kick things off with an inside look at who's behind Rafay.
Abstract
Teleco operators already own many of the ingredients needed for the AI era: facilities, power, networks, distributed infrastructure, enterprise relationships, and increasingly GPU capacity. Franck Jonas, Head of Telco, France, NVIDIA, explores how those assets can evolve into a distributed AI infrastructure platform—and what telcos need to add to turn infrastructure advantage into AI services and revenue.

Abstract
The gap between an AI factory design and a running, revenue-generating system is measured in months, and every one of those months is GPU-hour revenue lost on infrastructure already paid for. This session covers how the right systems integrator takes an operator from design to deployed across the full stack, getting them to first GPU-hour revenue sooner. It pinpoints where deployment stalls, and how the best operators avoid it.
Presented by World Wide Technology.

Connect with peers, continue conversations, make new connections, or meet with attendees you’ve pre-booked time with.
Secure and sovereign AI starts with a validated foundation. Cisco presents on the compute, fabric, and validated architectures that let operators serve regulated and sovereign demand.
He came to the role from the other side of the table. Before rejoining Cisco he spent several years inside a large telco leading finance transformation for the CFO, running the financial assurance organisation across a substantial operating £8.5Bn P&L and, in parallel, transforming the operating model to unify data and process assurance across the business. He led a team of 40 data scientists and a wider programme of 100 people across 15 departments: data assurance linked directly to the P&L, with machine learning and AI driving measurable efficiency across the order-to-cash cycle. That experience taught him where enterprise AI actually breaks. Not in the models, but in the structural layer beneath them: whether the data can be trusted, whether the processes hold, whether the infrastructure can carry production AI securely and at scale.
It is why he stands behind the structural capabilities of the Cisco platform today. He speaks about it as someone who was the customer, who hit those foundational problems first-hand, and who sees in Cisco's networking, data and security layer the capabilities enterprises genuinely need to make AI work in production. Earlier in his career Jon trained as an engineer, founded and exited a venture-backed SaaS company, and led the incubation of Cisco's collaboration cloud go-to-market across EMEA, growing the business beyond $1BN. He writes and speaks from experience, with a particular passion for sovereign AI infrastructure and what it takes to make AI commercially accountable.

Abstract
Neysa's Karan Kirpalani on building a business that beat the market on margins when few believed it could, and the discipline that turned into real scale. A candid read on what superior unit economics look like and how they compound.
Karan is recognized for driving technological innovation, managing product lines' P&L, and collaborating with major consulting and technology partners to enhance sales efforts in medium and large enterprises. He is also an active thought leader in the tech community, sharing insights on AI infrastructure and cloud strategies through various platforms. Notably, he has discussed the challenges of AI project implementations and the importance of AI-ready data centers in industry forums and publications.

Connect with peers, continue conversations, make new connections, or meet with attendees you’ve pre-booked time with.
Enjoy a great lunch while connecting with peers and industry leaders. Compare notes on the morning’s key takeaways, continue the conversations sparked by the sessions, and recharge for an engaging afternoon ahead.
Abstract
As operators scale from a single model to many, inference optimization becomes the key to margins. NVIDIA shares its perspective on the future of the token factory, exploring how to accelerate LLMs while driving down costs.
Presented by NVIDIA
The distance between hardware and a production AI business is where operators lose time and money.
AI's growth is widely seen as capped by power. The real constraint is where that power sits. Data center demand is set to outpace new grid capacity through 2030, and inference needs to run close to users to perform. The solution to both already exists in the ground: distributed, permitted power behind EV charging infrastructure in dense urban locations, waiting to be utilized.
Presented by Xeal.
Abstract
Most sovereign AI conversations focus on supply — GPUs, datacentres, national clouds. This session flips the lens to demand: what does a government actually need before it can use the capacity being built for it?
Farul’s work sits at the intersection of technology strategy and execution, covering infrastructure, platform architecture, product development, and AI deployment. He also works with government, academia, and industry on responsible AI adoption, governance, digital sovereignty, and strengthening Malaysia’s long-term AI capabilities.

Abstract
Rafay works across neocloud operators, enterprise buyers, and the partner ecosystem that connects them.
Mohan and Hemanth share what that work is surfacing: which neoclouds, telco and enterprise requirements come up in nearly every deal, where operators consistently underestimate the effort, how monetization models are shifting from hourly GPU rental toward consumption-based consumption, and what partners are asking Rafay to own versus build with them.
Haseeb closes on the shared playbook behind building AI as a business and, for many, an act of sovereignty.
Connect with peers, continue conversations, make new connections, or meet with attendees you’ve pre-booked time with.
