Latest Interviews

NextSilicon & More Than Moore. The Silicon Revolution: Why the Future of Hardware is Software

Next Silicon introduces a reconfigurable compute platform that utilizes a smart runtime algorithm to dynamically adjust chip hardware within 20 nanoseconds for CPU, GPU, and AI workloads without requiring code rewrites. This "software-aware hardware" targets the most computationally intensive code segments to deliver projected performance speedups of 4x to 10x while consuming only a quarter of the energy used by static ASICs. By supporting established languages and frameworks like CUDA and PyTorch, the system aims to decouple hardware efficiency from the need to pre-anticipate future algorithmic shifts.

A Chat with AI21 Labs: Building Custom AI Systems That Actually Work From Hype to Business Impact

Yuval, Asaf

AI21, a foundational LLM builder, highlights the persistent gap between individual AI experimentation and institutional production adoption, driven by "ChatGPT Syndrome" and fear of missing out. The company identifies a critical "underground movement" where employees innovate silently due to a lack of official guidelines, resulting in disjointed pilots and missed collective wisdom. To escape this paralysis, successful deployment requires multidisciplinary teams to map granular workflows and align stakeholders on clear KPIs, transforming informal usage into a structured, high-value organizational framework.

Roshn Group & Enhance Ventures: AI Meets Real Estate: Hype or Real Value

Robert, Yazeed Al-Ghamdi, Aristide, Maria Paula

Saudi Arabia's Roshan giga-project, mandated to deliver 400,000 housing units under Vision 2030, is leveraging Google Cloud AI and a proprietary digital sales platform to generate over 6 billion SAR in revenue while reducing land selection time by 40%. This hybrid transaction model, which integrates global KYC protocols and visualizes amenities for remote buyers, was further accelerated by the region's largest PropTech hackathon that sourced external solutions for operational bottlenecks. Driven by a young demographic and supported by upcoming infrastructure projects, the organization plans to launch a corporate venture capital arm and export its technology stack to international markets within the decade.

Neo4j, Eurazeo Growth: Connected Intelligence Why Graph Technology Is the Missing Link in GenAI

Romain, Philippe

Eurasio Fund identified graph technology as essential for bridging the gap between GenAI prototypes and production environments, a strategy validated by Gartner's finding that knowledge graphs increase success rates by 80%. Major enterprises like Klarna now leverage this approach to eliminate thousands of legacy applications and accelerate processing speeds by 1,000 times, while Neo4j serves as the underlying infrastructure for 84 Fortune 100 companies. The industry further solidified this shift in 2023 when the ISO standardized the Cypher query language as GQL, cementing graph databases as a critical component for explainability and security in complex AI systems.

Stephane Kasriel, Meta Fundamental AI Research (FAIR): Frontier AI From Research to Production

Stephane Kasriel, Stefan, Stephen Beyrer, Romain Guyett

Stefan Beirer leads Meta's Fundamental AI Research division in a $60 billion infrastructure expansion aimed at transitioning AI from academic theory to consumer-scale products through a Bell Labs-style model. This strategy leverages the open-source Llama ecosystem and real-time emotional avatars to drive global adoption while refuting immediate mass displacement fears by citing current hardware limitations. Beirer projects that massive compute investments will drive inference costs down by up to 1,000x within two years, enabling "always-on" devices and accelerating a decade of technological progress comparable to the first industrial revolution.

Ronan Harris (Snap Inc): Bringing AI into the world through augmented reality ft. Bloomberg News

Ronan Harris, Benoit

Snap plans to launch new consumer Spectacles next year while leveraging a platform strategy that currently supports 8 billion daily augmented reality lens uses among its 900 million users. The company is differentiating itself from hardware-focused competitors by fostering a developer ecosystem of 400,000 creators and integrating AI partners like Google and OpenAI to drive both ad targeting and third-party experiences. Despite facing stock declines and layoffs, Snap's leadership argues that adopting machine learning and resisting strict age bans are essential for maintaining market relevance against major global competitors.

Mark Carney on dealing with Donald Trump

Mark Carney, Donald Trump, Zanny Minton Beddoes, Prime Minister Carney, Sani

Prime Minister Carney attributes Canada's recent economic stagnation to persistent tariff uncertainty while characterizing U.S. trade policy as a coercive "mafioso" strategy driven by President Trump. In response to this geopolitical rupture, Ottawa is pivoting toward strategic sovereignty by bolstering Arctic defense and cultivating a "variable geometry" of trade alliances to reduce reliance on the American market. This dual approach acknowledges a permanent shift toward high-tariff U.S. protectionism, compelling Canada to prioritize self-sufficiency and diversify global partnerships beyond traditional transatlantic frameworks.

Paul Copplestone (SupaBase) & Abel Samot (Commit): Open Source Backend for the AI Age

Paul Copplestone, Abel Samot, DAN GALPIN, FRANCESC CAMPOY, ALEXIS MOUSSINE

Supabase has achieved a $2 billion valuation following a $200 million funding round, driven by a user base that tripled to over 3 million developers and a strategy positioning the open-source platform as a default modern web backend. Founder Paul Copplestone highlights rapid adoption fueled by AI-assisted "vibe coding" tools while the company pursues enterprise growth through "configuration over customization" and upcoming infrastructure breakthroughs like the Multigress sharding solution. Operating as a fully remote organization of 140 specialists, the firm continues scaling by launching Pico instances for zero-to-one applications and recruiting senior talent to support petabyte-scale database capabilities.

A fireside Chat with May Habib, CEO & Co Founder of Writer ft. Rana Yared, GP, Balderton Capital

May Habib, Rana Yared

Writer is an enterprise AI company accelerating the transition to "Era 3.0" agentic systems by deploying vertically specific, revenue-impacting solutions that integrate with legacy infrastructure while maintaining full-stack control over security and reliability. The organization addresses adoption barriers like traceability and scalability through a frugal, synthetic-data-driven architecture that enables rapid model iteration and distinct competitive moats against infrastructure providers. By structuring implementation around 30-to-365-day value increments, Writer facilitates deep corporate transformation from efficiency tools to autonomous agents capable of executing complex, high-stakes business workflows.

Together AI & SemiAnalysis: In Conversation Together AI's Vision For the Future of AI Infrastructure

Vipul Vaid-Prakash, Dylan Patel

Founded in late 2022, Together AI challenges hyperscaler dominance by reengineering the cloud stack to achieve 75% maximum FLOP utilization on H100 GPUs, significantly outpacing the 50% industry standard. Led by chief scientist Tri Dao and the ClusterMax initiative, the firm drives down inference costs for massive models like DeepSeek from $8 to 30 cents per million tokens through architectural innovations and Reinforcement Learning frameworks. These technical advancements enable new MoE-based workloads and position the company to define emerging optimization surfaces before new hardware generations are fully adopted.

A Fireside Chat with Nicolas Dessaigne (YC-Combinator) & Jean de la Rochebrochard (Kima Ventures)

Nicolas Dessaigne, Jean de la Rochebrochard, Nicolas Tassin

Kima Ventures' Jean de la Rochebrard and Y Combinator partner Nicolas Tassin discussed their investment strategies, emphasizing that AI has shifted from a distinct category to a de facto standard comparable to cloud computing. The investors outlined distinct regional dynamics, noting that US founders leverage superior risk tolerance and speed while European entrepreneurs must overcome cultural humility to access the capital pools required for billion-dollar growth. They concluded that future market leaders will likely emerge in vertical applications within "boring" industries rather than horizontal platforms, provided founders prioritize authentic execution over hype and build defensible data moats.

A fireside Chat with Paddy Srinivasan, CEO of DigitalOcean ft. Semafor

Paddy Srinivasan, Patti, Reid

DigitalOcean CEO Patti and Reid Hoffman discuss the platform's role in serving 3 million developers by providing specialized AI infrastructure for both native startups and traditional firms integrating large language models. The conversation highlights critical global bottlenecks, including power constraints necessitating nuclear energy and a talent war driving new hiring protocols for AI automation engineers. Hoffmann forecasts a massive economic expansion where AI increases the technology sector's share of global GDP to 25% within a decade, alongside a predicted dispersion of value beyond current market leaders.

Ioannis Antonoglou (Reflection AI): Building The Next Superintelligence After AlphaGo

Ioannis Antonoglou

Reflexion AI co-founder Yiannis and his team argue that AGI requires no new scientific breakthroughs, but rather the assembly of existing knowledge to solve the foundational "root node" problem of autonomous coding agents. Differentiating from well-funded competitors like DeepMind and OpenAI, the startup prioritizes a ruthless focus on software development while shifting training paradigms toward reinforcement learning in simulated environments. The founders also analyze the global landscape, attributing Europe's lag to market fragmentation and regulation, while asserting that mission alignment offers a superior talent acquisition strategy to massive signing bonuses.

A Fireside Chat with Luc Julia, Renault & Matthieu de Chanville, Shift for Good

Luc Julia, Matthieu de Chanville

Luc Beaulieu, Renault's Chief Scientific Officer, highlights the critical lag between century-old automakers and agile startups regarding AI adoption, noting that legacy software cycles remain incompatible with the rapid hardware refresh rates required for software-defined vehicles. While dismissing the feasibility of Level 5 autonomy and humanoids, Beaulieu champions specialized "physical AI," geo-fenced Level 4 robot taxis, and virtual sensor technologies to enhance manufacturing efficiency and reduce environmental impacts. These insights are further reinforced by the "Software Republic" consortium, which includes Renault and Thales, as it advances a vision where artificial intelligence serves as a specialized tool to augment human capability rather than replace it.

An AI conversation with Prashanth Chandrasekar, CEO of Stack OverFlow

Prashanth Chandrasekar, Henry Shi, Prashant

Stack Overflow has pivoted from a traditional Q&A hub into a revenue-generating knowledge intelligence layer by licensing its data to major AI providers like Google and Microsoft while launching enterprise solutions for corporate workflows. This strategic shift addresses a 64% year-over-year traffic decline caused by user migration to AI search, transforming trust deficits among senior developers into a premium verification service rather than a code-generation competitor. Despite risks to junior developer roles, the company remains profitable and is expanding its talent pipeline through certification challenges and asynchronous human-to-human chat features.