Anthropic's Generational Run, OpenAI Panics, AI Moats, Meta Loses Major Lawsuits
Silicon Valley is fracturing into rival camps as Anthropic's enterprise blitz collides with OpenAI's consumer dominance — and both face existential questions about durability. Are we witnessing the dawn of infinite abundance that will collapse every business model, or just the latest SaaS cycle dressed in agentic clothing? Meanwhile, Meta faces a legal reckoning that could redefine corporate liability in the digital age.
Points clés
Anthropic has seized enterprise dominance through focused execution on coding as the gateway to IT budgets, while OpenAI retains consumer brand but is losing market share and strategic focus.
The «SaaS apocalypse» signals a deeper market revaluation: investors are demanding cash-flow certainty over growth as AI erodes the durability of software moats.
Meta lost two major verdicts in two days — one for child exploitation ($375M), one for designing addictive products — opening the door to «death by a thousand cuts» tort litigation despite Section 230 protections.
The real AI battleground is shifting from model performance to industrial deployment: private equity, enterprise rollups, and change-management services will capture value as businesses struggle to operationalize LLMs.
Consumer AI will not remain free — premium subscriptions for AI services could exceed Netflix and Spotify, reaching hundreds of millions of users willing to pay $20–80/month for ultra-valuable meta-services.
En bref
The AI race is splitting: Anthropic owns the enterprise with superior coding tools, OpenAI still commands consumer mindshare but is pivoting frantically, and both face the hardest question in capitalism: what is anything worth if super-intelligence makes all advantages temporary?
Anthropic's Enterprise Dominance
Anthropic's focused bet on coding unlocked rapid enterprise revenue growth.
Anthropic made a strategic wager on coding as the breakout use case — whether for AGI ideology or business pragmatism — and it paid off spectacularly. Coding became the gateway into enterprise IT budgets, enabling rapid revenue scale. The company extended from Claude Code to Claude Co-work (generating PowerPoints and spreadsheets via code generation) and now to computer-use agents that mimic OpenClaw functionality. They added $6 billion in annual run rate in February alone.
Jensen Huang called Opus 4.6 an «inflection point» and the first truly agentic model. Michael Dell said it hit a productivity threshold unseen before. Sacks praised the product quality while noting philosophical objections to Anthropic's regulatory-capture strategy — advocating for permissioning regimes in Washington that could entrench incumbents. Despite not being Trump administration donors and maintaining critical public stances, Anthropic has been treated no differently in procurement. The consensus: they're firing on all cylinders technically, even if their ideological positioning remains controversial.
OpenAI's Strategic Whiplash
OpenAI retains consumer brand dominance but is losing share and focus.
Consumer Market Share Erosion OpenAI created the category in 2023 with 100% share, dropped to 85% in 2024, and sits at 75% in 2025. While absolute usage grows, competitors are gaining ground.
Apple, Meta, Windows: MIA but Looming Three major players — Apple, Meta, Microsoft — remain underrepresented in consumer AI. If they capture even modest share, OpenAI could fall below 50% market dominance.
Sora Video App Cancelled OpenAI shut down Sora, cancelling a billion-dollar Disney licensing deal and integration into Disney Plus. Signals retreat from side projects.
Pivot to Enterprise OpenAI is now chasing Anthropic down the enterprise path, offering private equity firms a guaranteed 17.5% minimum return on joint ventures to ease AI deployment costs.
The SaaS Apocalypse: A Valuation Reckoning
AI uncertainty is collapsing SaaS multiples as markets question durability.
Chamath presented a stark chart: SaaS companies once trading at 100 years of free cash flow (Snowflake in 2023) have been cut in half. ServiceNow, Workday — all re-rated down. The market is asking: if super-intelligence is coming, how durable are any cash flows? Won't all companies be disrupted every five to six years? The rational employee response: «I don't want your equity, give me more cash.» This is a societal question about capital markets, not just a sector rotation.
Meanwhile, the Mag 6 (Apple, Microsoft, Meta, Alphabet) have walked valuations *up*, signaling investor belief in monopolistic durability forever. Nvidia — despite $200B profits and unmatched execution — is being treated like a SaaS company. The divergence reflects a fundamental revaluation: free cash flow certainty is the new moat. Enterprises are already asking for «strangulation as a service» — AI shims that eliminate complex UIs entirely, letting users simply describe tasks and bypass entire product surfaces.
Chamath on Super-Intelligence and Market Value
If infinite abundance is real, what is anything worth in year 10?
“What if there's this super intelligence on the horizon? I think it's fair to ask the question, what is anything worth? Won't all companies be disrupted? And if you were faced with that problem in the public markets, how would you react? The canary in the coal mine are the SaaS stocks.”
Meta's Legal Reckoning
The Tort Tax Debate: Corporate vs. Parental Responsibility
Should companies be liable for harm, or do parents bear responsibility?
Consumer AI: Free or Premium?
The debate: will consumer AI be ad-supported or subscription-driven?
Calacanis argues consumer queries will be free — Apple, Google, Meta will subsidize with ads or ecosystem lock-in. ChatGPT's paid tier (roughly 50 million subscribers out of 900 million+ users, ~5%) mirrors typical freemium conversion. Friedberg counters: AI will be the most valuable meta-service consumers have ever seen. Spotify has 290M paid subscribers, Netflix 325M. Why not 300–500M people paying $20–80/month for AI that books travel, manages email, tracks calendars? It's more valuable than cable or cell service.
Sacks notes B2B is stickier and more lucrative for investors, but consumer businesses like Meta and Google are the most valuable in the world. The outcome depends on how aggressively Google bridges its ad network to AI chat — and whether Apple leverages its ecosystem to make AI indispensable. One certainty: the iPhone was the platform for the app economy; consumer AI apps could become the platform for an embedded-services economy where advertisers pay to appear inside agents.
Key Data Points
Critical numbers shaping the AI and social media landscape.
PCAST: Silicon Valley Goes to Washington
Trump's science council stacked with builders, not just academics.
President Trump appointed David Sacks to co-chair the President's Council of Advisers on Science and Technology (PCAST) alongside OSTP Director Michael Kratsios. The roster includes Marc Andreessen, Sergey Brin, Michael Dell, Larry Ellison, David Friedberg, Jensen Huang, Lisa Su, and Mark Zuckerberg. Sacks noted the council is heavier on «doers and builders» than prior administrations — people who invented the GPU, the browser, quantum computing advances.
Friedberg emphasized the existential stakes: 10 years ago, China published 50% as many scientific papers as the U.S.; last year, 50% more. The race is no longer just discovery — it's industrial deployment. This is why builders matter. PCAST can appoint up to 24 members; nine slots remain open for additional expertise. The council will advise on AI, quantum computing, nuclear power, advanced semiconductors, and biotech — domains where America's lead is no longer guaranteed.
Final Thought: The Simulation Feeling
AI tools now manifest ideas instantly, collapsing time and cognition.
Final Thought: The Simulation Feeling
Friedberg described building a Chrome extension for a 15-year-old idea (annotated.com) in a single weekend using vibe-coding. Chamath's team doubled click-through rates overnight using auto-research. The feeling is disorienting: ideas that would have taken man-months and teams now materialize in days. It's the replicator from Star Trek — «Earl Grey tea, hot» — applied to business. Time is collapsing. Every founder now experiences the same vertigo: what do I do with all this suddenly accessible capacity?
Personnes
Glossaire
Avertissement : Ceci est un résumé généré par IA d'une vidéo YouTube à des fins éducatives et de référence. Il ne constitue pas un conseil en investissement, financier ou juridique. Vérifiez toujours les informations auprès des sources originales avant de prendre des décisions. TubeReads n'est pas affilié au créateur de contenu.