Latest Interviews
Showing 346–360 of 9,611 transcripts.
Clear all filtersFrontier Labs Threatened by Kimi? Should the US Ban Chinese Open-Source Models & Stripe Buys PayPal
Jason Lemkin, Rory O'Driscoll, Harry Stebbings
China's Moonshot AI and Alibaba recently launched near-frontier open-weight models like Kimi and Qwen that are rapidly capturing significant market share and challenging US dominance despite a six-to-nine-month technology gap. While political friction over "AI communism" and data sovereignty risks creates regulatory uncertainty, aggressive engineering and massive capital injection are accelerating the convergence of Chinese and US model performance. Simultaneously, the sector is witnessing a strategic shift toward vertical integration and specialized micro-models, as investors anticipate that open-weight economics will force legacy US frontier providers to compete with inference costs up to 80% lower.
- RAISE Summit20 min
Fireside Chat with William Falcon & Ozan Kaya of Lightning AI | RAISE Summit 2026
William Falcon, Ozan Kaya, Olivier Huez, Will
Following a rapid expansion to 160 employees, Voltage Park merged with Lightning.ai to transition from a GPU-as-a-service provider into a full-stack platform serving Gen AI enterprises. This combination integrates 24,000 NVIDIA H100 GPUs across six data centers with Will's software tooling to help clients own their open-source models rather than relying on hyperscalers. The strategic alliance addresses the market gap for dedicated compute, prioritizing high-uptime infrastructure and digital sovereignty for customers like Cursor.
- RAISE Summit19 min
Fireside Chat with Tess Munsie, Chief of Staff at Fundamental | RAISE Summit 2026
Tess Munsie, Nina dos Santos, Tess Muncy
Fundamental has emerged from stealth with a $1.45 billion valuation to deploy Nexus, a Large Tabular Model specifically engineered to solve the "blind spot" where traditional LLMs fail to process structured enterprise data with deterministic precision. By utilizing a "bring the model to the data" architecture within customer VPCs, the platform enables industries like energy and shipping to transform decades of trapped records into immediate operational insights while maintaining strict data sovereignty. This approach reduces forecasting timelines from months to hours and delivers a generalized solution that replaces brittle, task-specific pipelines across sectors ranging from CPG to sports analytics.
- RAISE Summit19 min
Building Africa's AI Factory: Cassava, Rafay and Sovereign AI | RAISE Summit 2026
Cassava Technologies, a digital solution provider serving over 40 countries, has secured direct investment from NVIDIA and Google to deploy a GPU-rich compute infrastructure across Africa, Latin America, and India. The company distinguishes itself through a custom orchestration layer that democratizes access to AI for students and startups, enabling the development of localized language models and critical sector applications in agriculture and healthcare. With a strategic focus on expanding dark fiber connectivity and green energy integration, Cassava aims to drive economic growth and job creation in emerging markets over the next five years.
- RAISE Summit18 min
Zero Flaps, Zero Excuses: Network Reliability at AI Scale | Credo x TensorWave | RAISE 2026
Credo SVP Phil Kuhman and Tensorwave Co-founder Jeff Tatarchuk discussed the critical role of high-reliability networking in scaling AMD-exclusive GPU clusters, citing a deployment of 8,192 MI325 GPUs supported by Tensorwave's two-gigawatt North American infrastructure. The dialogue highlighted Credo's Zero-Flap transceiver technology, which utilizes rigorous thermal hardening and in-band telemetry to reduce deployment cycles from six weeks to one and eliminate the costly downtime caused by link failures in hyperscale environments like Oracle's. By driving the industry toward a standardized "Purple Cables" architecture that combines copper for short-range reliability with advanced optics for long-haul connectivity, the companies aim to transform networking from a bottleneck into a table-stakes foundation for rapid AI infrastructure revenue.
- RAISE Summit19 min
Inside the AI Inference Cluster: Measuring What Matters | Mansour Karam, Aria Networks | RAISE 2026
The AI infrastructure sector is pivoting from homogeneous general-purpose systems to specialized architectures that optimize distinct model training and serving economics, with high-throughput inference offering potential cost savings of up to 100x. This shift demands deep networking solutions capable of managing dynamic, unpredictable traffic patterns and microsecond-scale latency requirements for complex agent workflows. Leading this transition, ARIA proposes a multi-layered AI agent architecture that integrates full-stack telemetry to dynamically coordinate hardware, software, and network operations across immediate physical reactions and long-term strategic planning.
- RAISE Summit18 min
Where We're Going We Don't Need Keyboards, Just Voice | Neil Zeghidour, Gradium | RAISE 2026
Gradium, a spin-off of the frontier research lab Qtize, is commercializing full-duplex voice AI models like "Moshi" to replace traditional keyboard interfaces with natural, real-time human-machine interaction. By utilizing single audio-to-audio architectures that eliminate turn-taking latency, the Paris-based company enables seamless agentic tasks ranging from customer support to industrial robotics in noisy environments. With recent funding to expand into San Francisco and open-source model access, Gradium aims to establish voice as the primary operating system for always-on, asynchronous agent networks.
- RAISE Summit20 min
Why the Next Marketing Giant Will Be Hyperlocal, Not Global | HubKonnect | RAISE Summit 2026
Michael outlines a critical market shift where enterprise marketing transitions from national campaigns to hyperlocal activation driven by a centralized intelligence layer that compounds data across millions of location-specific points. This operational reset requires C-suite alignment to replace obsolete agency models with an "Ecosystem as a Service" approach, ensuring data reciprocity and transforming marketers into real-time profit center drivers using Hyperlocal Intelligence. Ultimately, the differentiator for future growth lies not in isolated SaaS tools but in "better operators" who leverage proprietary datasets to execute culture shifts that empower rapid, data-informed decision-making at the individual unit level.
- RAISE Summit19 min
What's New for Startups with Google DeepMind? | Omar Sanseviero | RAISE Summit 2026
Omar Sanseviero of Google DeepMind outlined the strategic expansion of the Gemini ecosystem, highlighting the release of Gemini 3.5 Flash, the imminent Gemini 2.5 Pro, and advanced GenMedia models like NanoBanana 2 Lite for rapid image generation. The presentation detailed critical infrastructure updates including the Gemma 4 open model family, which achieved over 250 million downloads on consumer hardware, alongside new real-time APIs for low-latency voice translation and live visual analysis. These advancements collectively aim to accelerate the transition from prompt to production by empowering developers to build complex agentic workflows and specialized applications in sectors ranging from robotics to environmental monitoring.
- RAISE Summit18 min
Operating AI Infrastructure at Multi-Model, Multi-Cloud Scale | RAISE Summit 2026
Lukas Gentele, Bola Malek, Nikita Vdovushkin
Nebius and Base10 are redefining the AI infrastructure landscape by pivoting from raw compute provision to managed "Token Factory" services that prioritize inference efficiency and multi-cloud orchestration across dozens of providers. While Base10 aggregates demand across 90+ regions to ensure redundancy and data sovereignty, Nebius leverages vertical hardware integration to optimize cost-to-performance ratios, together supporting a market shift toward hybrid open-source and proprietary model strategies. This collaboration and competitive tension address the growing enterprise demand for transparent, token-based pricing and localized hosting solutions that balance geopolitical constraints with the need for scalable, specialized AI deployment.
- RAISE Summit20 min
From Software to Hardware: The Next Wave of AI | Atomico x PhysicsX | RAISE Summit 2026
Physics X recently secured a $300 million Series C led by Temasek, raising its valuation to $2.4 billion to accelerate its expansion into Singapore and the broader Asian market. Founded by ex-Formula 1 engineers, the company utilizes AI-native physics models to deliver simulation speeds up to one million times faster than traditional methods for clients in aerospace, automotive, and mining sectors. To address complex engineering challenges in defense and energy transition, the firm operates with a forward-deployed team model while transitioning its platform to support self-service graphical interfaces and automated workflows.
- RAISE Summit19 min
Digital Workers for the Physical Economy: AI Agents in Global Supply Chains | Magentic | RAISE 2026
Magentic positions itself as a leader in the physical AI economy by leveraging entity resolution technology to automate procurement and supply chain operations for Fortune 500 clients. The company differentiates through an outcome-based pricing model that charges only upon realizing tangible savings, having already delivered rapid ROI for major industries ranging from electronics to brewing. By converting legacy data into actionable insights, Magentic enables physical enterprises to replicate the growth velocity of software firms while preparing procurement teams to expand their strategic roles over the next five years.
- RAISE Summit20 min
Running Secure Agents on Multi-Tenant Sovereign AI Infrastructure | Protopia AI x HPE | RAISE 2026
Eiman Ebrahimi, María Ridruejo, Maria Redrojo, Imon Ibrahini
Protopia AI and HPE AI Services have introduced the Stained Glass Transform, a privacy-preserving technology that converts sensitive data into randomized, non-reversible representations before they enter multi-tenant AI factories. This innovation, integrated through the SafeClaw subagent architecture, enables agentic workflows to run on shared infrastructure without compromising security or suffering the GPU underutilization associated with traditional hardware isolation. Consequently, sovereign providers and neoclouds can unlock significant revenue by processing high-volume sensitive workloads with near-baseline latency, addressing the critical gap between data ownership and economic efficiency in modern AI deployment.
- RAISE Summit20 min
Benchmarks vs. Reality: Lessons from 750 Trillion Tokens | Chris Clark, OpenRouter | RAISE 2026
OpenRouter leverages empirical usage data from over 80 cloud providers to highlight a critical gap between published AI pricing and real-world inference costs, driven by variable caching efficiencies and tool call reliability. The analysis reveals that while open-weight Chinese models currently lag US frontier capabilities by four to six months, successful adoption depends less on raw intelligence rankings than on specific optimization for agentic workflows and stable inference. Consequently, industry leaders are advised to prioritize production-based validation over theoretical benchmarks to navigate shifting cost structures where inference expenses are rising to rival headcount as a primary operating burden.
- RAISE Summit17 min
Headline AI Europe 100: 2026 Megatrends & Next Winners | Headline | RAISE Summit 2026
Astrid Moullé-Berteaux, Jonathan Userovici, John
Headline Ventures has introduced a new sector-agnostic classification framework that categorizes AI startups into six distinct "breeds" based on economic models rather than traditional industry verticals. Simultaneously, the firm announced the second edition of the "AI Europe 100" list, highlighting 100 European startups valued below $1 billion while analyzing regional strengths in deep research and gaps in applied AI sectors. This initiative aims to align founders with investment criteria by emphasizing specific operational metrics and scalability potential across the diverse ecosystem.