**Duration:** 60 minutes (Dr. Saif departs at 45 mins) **Key theme:** The thesis areas we're co-building and investing in at [The Utopia Studio](https://www.utopia-studio.co/): **[infrastructure intelligence](https://www.utopia-studio.co/infrastructure-intelligence) and [sovereign systems](https://www.utopia-studio.co/sovereign-systems)**. ## The Panelists **[Dr. Saif M. Al-Kuwari](https://www.linkedin.com/in/smalkuwari/)** | Associate Professor, **[HBKU](https://www.hbku.edu.qa/en)** (Qatar) Lead for Qatar's quantum technology efforts. Deep expertise in quantum communications and security. Bridges theory to national capability. **[James Walsh](https://www.linkedin.com/in/james-walsh-15b40591/)** | Managing Director, **[Sytronix](https://www.sytronix.co.uk/about)** (UK) Designs best-performing hardware systems and servers globally. Built bespoke configuration for The Utopia Studio. **[Alexandra Coleman](https://www.linkedin.com/in/alexandra-coleman20/)** | Founder & CEO, **[Azraq](https://azraqdc.ai/)** (Middle East) Full-lifecycle risk management platform for digital infrastructure. Structural engineer turned infrastructure intelligence founder. Fellow at The Utopia Studio. **[Taha BELMEJDOUB](https://www.linkedin.com/in/tahabel/)** | **[Sia Partners](https://www.sia-partners.com/en)** (Middle East) AI applications specialist. Deep knowledge in commercialization and adoption. Sia operates one of the strongest [[AI Agents Stack|Agentic AI labs]] in their toolkit. --- ## Thread 1: Quantum and the Next Frontier (Dr. Saif, 15 mins) Everyone wants to be invested today where they should have been 5 years ago. That's where quantum is. See: [[Themes shaping 2025#4. Quantum Tech continues rapid advancements]] **Questions:** 1. What is the significance of quantum technologies? Where does the hype end and reality begin? 2. Walk us through your work in quantum communications and security. What's the [[What is the quantum threat|quantum threat]] that organizations should prepare for? 3. How do quantum and AI converge? The hardware angle, the software angle. 4. Companies to watch and why? 5. The sovereignty angle: with national quantum programs worldwide, what does quantum mean for Qatar's strategic positioning? 6. What breakthrough are you most excited about seeing unfold? **Context to weave in:** - [[Quantum 101]]: Quantum is non-deterministic computing. Adding a qubit doubles computational power. The ratio of physical qubits to logical qubits doing actual work is the key constraint. - [[What is the quantum threat]]: Asymmetric cryptography (SSL, TLS) underpins digital signatures, identity authentication, key transport. Fault-tolerant quantum computers could break these in hours or seconds. - [[Why start preparing now for the Quantum Threat]]: "Harvest now, decrypt later" means sensitive data captured today becomes vulnerable when quantum arrives. - [[Recommended quantum readiness best practices]]: Organizations should transition to quantum-resistant cryptography by 2025-2026. Start with cryptographic inventory, then plan migration. - [[Post Quantum Cryptography]] and [[NIST PQC Standards]]: The standards are here. [[PQC Roadmap Questions to ask Vendors]] gives organizations the right questions to ask. - [[Quantum Security MOC]] and [[Quantum Security Thesis]]: The commercialization pathway for quantum-secure networks and [[Quantum Readiness Program]] implementation. - [[Likely road to Quantum Value]]: NISQ value exists for scientific exploration, but commercially relevant applications need [[Quantum Error Correction 101|error correction]] and fault-tolerant quantum computers. - [[Quantum Use Cases MoC]]: From [[Quantum x Energy]] to [[Quantum x Pharma]] to [[Quantum x Quant Finance]], the application map is broad. - [[Public Quantum Tech Ecosystem Companies]] and [[Quantum Ventures]]: Who to watch in the space. --- ## Thread 2: Hardware Systems and Data Centers (James, 15 mins) Global AI investment now dwarfs Manhattan Project, Apollo, and the New Deal combined to become one of the largest sustained investment programmes in human history. Compute is becoming an asset class. The cost of compute is the cost of software innovation. **Questions:** 1. Your story: how did Sytronix evolve into designing the world's best-performing hardware systems? 2. For someone building AI infra, what actually matters? Application/model fit, scalability, ownership. 3. If you're an AI startup and 50% of costs go to compute, how should they think about price-performance, ownership, sovereignty? 4. What breakthrough in hardware are you most excited about? **Context to weave in:** - [[The infrastructure layer and AI capex]]: AI infrastructure CapEx absorbing ~5% of US GDP, comparable to fiber buildout of early 2000s. Real value accrues to prudent players who harness infrastructure, not those chasing the boom. - [[AI Capex Super-Cycle]]: [[Hyperscalers]] committing $400B+ pursuing compute supremacy. Individually rational, collectively irrational. If Google offers Gemini for free, competitors may never grow revenues to compensate for their buildouts. - [[Data Center MoC]]: The full stack from [[Data Centre First Principles]] to [[Data Center Tiers]] to [[Data Centre Metrics]]. - [[Heat is the new constraint]]: AI chips now exceed 1000W [[Thermal Design Power - TDP]]. Air cooling can't keep up. Liquid is mandatory. [[Direct-to-Chip Cooling]] handles 1000-1600W. Beyond that, [[Two-Phase Cooling]] or [[Immersion Cooling]]. Capital flows to atoms, not bits. - [[Cooling and Energy Efficiency Technologies]]: Over $1B in venture capital into cooling startups since 2023. Strategic acquirers paid 15-25x revenue multiples. - [[Puzzle of low data center utilisation]]: Widespread perception that data centers run 24/7 at max capacity is wrong. Load factor ≠ utilization rate ≠ server uptime. Hardware fragility at scale: Meta's Llama 3 training saw 466 job interruptions in 54 days across 16,000 H100s. - [[Modular Data Center Design Principles]] and [[Modular Data Center Opportunity in the Middle East]]: The regional angle. - [[The Watt-Bit Spread and the Future of AI Power Markets]]: The economics of power for AI. - [[SMRs x Data Center Opportunities]] and [[Geothermal Energy and the Future of Data Centers]]: Alternative power paths. - [[Needs of Hyperscaler Offtakers - Maximizing Compute Capacity at Scale and Speed]]: What do hyperscalers actually care about? - [[The AI Stack - Building Blocks]]: Infrastructure layer → Intelligence layer → Applications layer. Progress in any single layer drives advancements across the entire system. --- ## Thread 3: Infrastructure Risk Intelligence (Alexandra, 15 mins) [[Maintenance CapEx|Maintenance is the next big wave]]. Cost of generation is falling, cost of distribution and transmission is rising. Risk intelligence for investment and operational decisions is scarce. **Questions:** 1. Your journey from structural engineering to building a risk intelligence platform. What did you see that others missed? 2. The value of risk intelligence for the physical world. Why is quantifying physical risk so hard? 3. How do you keep pace when technology shifts constantly? World models, new sensing, AI. 4. Teach us something niche that makes us smarter walking out. **Context to weave in:** - [[Maintenance CapEx]]: $106 trillion infrastructure base by 2040 needs maintenance, regardless of economic cycles. The "boring" category is where the money is. - Operations and maintenance: 56.7% of all public infrastructure spending, up from 44.4% in 1970s. The shift from "building new" to "maintaining old" is already happening. - Maintenance spending is non-discretionary. Physical assets deteriorate without intervention. Deferred maintenance compounds at ~7% annually, eventually forcing catch-up spending regardless of interest rates. - Massive perception gap between "maintenance" (seen as low-tech, low-margin) and reality (cutting-edge sensors, AI, robotics, advanced materials). This arbitrage creates pricing power. - Technology enables new business models: sensors at lower SWaP-C than ever, AI models that ingest raw sensor streams, mobile connectivity in emerging markets. - [[Data Centers could make or break electricity affordability]]: Infrastructure failures have ripple effects through the system. --- ## Thread 4: AI Applications and Value Creation (Taha, 10 mins) [[model layer and above|Real value will accrue at the application layer]], not infrastructure. We need adoption, we need value-generating use cases. See: [[Themes shaping 2025#1. More AI Autonomous, Invisible and SaaS killer]] **Questions:** 1. Examples of AI applications creating real value for clients in the Middle East today. 2. State of AI adoption: how should a company looking to invest in AI actually spend their money? 3. Lesser-known, high-value areas for AI application that excite you. 4. Teach us something that makes us smarter walking out. **Context to weave in:** - [[The AI Stack - Building Blocks]]: Applications layer is where users interact. This is where transformation happens. - [[AI Agents Stack]]: Agent ecosystem developed significantly. Progress in memory, tool usage, secure execution. But frameworks vary wildly in state management, context window handling, memory approaches. The shift from LLMs to LLM agents requires new infrastructure. - [[Autonomous AI Agents - The rise, potential and challenges]]: Agents moving beyond assistance to independent action. Managing workflows, solving intricate problems, creating new efficiencies. - [[model layer and above]]: Closed vs. open models, verification as the bottleneck, AI's leverage on human labor. What is really being acquired isn't hardware but human leverage. Two engineers could outperform a hundred lawyers. - [[AI era Defensibility]]: Growth alone doesn't make you Google. Defensibility is your startup's ability to survive competitive pressure over time. Think like a castle builder: bailey (fast-moving outer defense) and motte (deep, harder-to-build core). - [[AI Verification]]: AI capability limited less by model size and more by what we can verify. For tasks with binary outcomes, AI excels. For subjective tasks, progress is slower. Errors compound: even 99% reliable agent falls to 60.5% accuracy after 50 sequential decisions. - Past cycles suggest investors overspend during euphoria. The gap between "deployed AI" and "valuable AI" is where the opportunity lives. --- ## Closing Hooks (if time permits) - "What would you tell your 25-year-old self about this space?" - "What's one myth in your domain that needs to die?" --- ## Related Notes **Quantum:** [[Quantum 101]] | [[Quantum Security MOC]] | [[What is the quantum threat]] | [[Why start preparing now for the Quantum Threat]] | [[Recommended quantum readiness best practices]] | [[Post Quantum Cryptography]] | [[NIST PQC Standards]] | [[Quantum Readiness Program]] | [[Quantum Error Correction 101]] | [[Likely road to Quantum Value]] | [[Quantum Use Cases MoC]] | [[Quantum Ventures]] | [[Public Quantum Tech Ecosystem Companies]] **Data Centers & Hardware:** [[Data Center MoC]] | [[Heat is the new constraint]] | [[Cooling and Energy Efficiency Technologies]] | [[The infrastructure layer and AI capex]] | [[AI Capex Super-Cycle]] | [[Hyperscalers]] | [[Puzzle of low data center utilisation]] | [[Modular Data Center Design Principles]] | [[The Watt-Bit Spread and the Future of AI Power Markets]] | [[SMRs x Data Center Opportunities]] | [[Data Centre Energy Demand]] **Infrastructure & Maintenance:** [[Maintenance CapEx]] | [[Predictive Maintenance in O&G]] | [[Data Centers could make or break electricity affordability]] **AI Applications:** [[The AI Stack - Building Blocks]] | [[AI Agents Stack]] | [[Autonomous AI Agents - The rise, potential and challenges]] | [[model layer and above]] | [[AI era Defensibility]] | [[AI Verification]] | [[Themes shaping 2025]] --- Tags: #deeptech #kp