AI never has a dull moment. My colleague Aman Kabeer and I recently recorded an engaging episode of The MAD Podcast, where we explored market observations, favorite trends, emerging stories, and future predictions. Watch the video here, and below are some highlights: Listen on: Spotify: https://tinyurl. com/yc6emzhc Apple Podcasts: https://tinyurl. com/4payud8p Current Market Highlights: * Record-breaking milestones: OpenAI achieved the largest VC round, Safe SuperIntelligence raised the biggest seed round at $1B, Character had the largest acqui-hire at $2. 7B, and xAI unveiled the biggest supercomputer, Colossus. * Sky-high valuations: NVIDIA’s P/E is at 65, much higher than the 20 average. Several startups have multi-billion valuations despite being pre-revenue or pre-product. Growth expectations are high, whether warranted or not. * Rapid growth for AI startups, outpacing previous SaaS companies (according to Stripe data). Even with worries about overbuilding AI infrastructure, recent earnings from Mag 7 companies indicate strong or "insane" demand for AI, though overall numbers remain modest by Mag 7 standards. * Uncertainty about AI research progress: GPT-5 may not deliver exponential improvements like its predecessors. Questions about LLM reasoning and the meaning of AGI/ASI persist. Despite this, optimistic views suggest current AI capabilities are sufficient for building great products and companies, even if progress slows. * The foundational/frontier LLM market has early leaders, yet it's unclear if long-term differentiation is achievable. The impact of Llama 4 and open-source projects on commercial API providers is in question.

There's still ample opportunity for specialized LLMs in various modalities (audio, video) or sectors (biotech, materials science, etc. ). * The AI tools layer (orchestration, evaluation, RAG) is intriguing but crowded and rapidly evolving. The AI consumer application sphere is wide open, akin to past paradigm shifts leading to dominant consumer companies. Who will create the Uber of Generative AI? * From a customer viewpoint, enterprise AI adoption is in its infancy, focusing on basic use cases: chat (GPT/Azure), search (Glean), code (GitHub Copilot), and user-friendly applications (Synthesia for video). For customized or industry-specific solutions, consultants are reaping the benefits. The next step: enterprises advancing from proofs of concept to production with bespoke use cases. * From a vendor angle, AI (intelligent overlays) is becoming the new SaaS (database overlays), with a frenetic pace of developments across use cases (AI for finance, sales, marketing, HR, etc. ) and verticals (law, accounting, banking, etc. ). New business models are emerging, such as "selling the work" instead of just selling software. * We’re heading into a brave new world. In enterprises, this might mean AI agents autonomously managing end-to-end tasks. Future enterprises may require fewer people, shifting job roles to coordinate between a few humans and many AI systems.