Hyperscalers burn cash to dominate AI's physical bottleneck

AI & Tech (2)

• Sacks notes OpenAI released GPT 5.5 Cyber, which matches Anthropic's Mythos in completing multi-step cyber attack simulations, and is likely the first such model commercially available due to OpenAI's superior compute capacity.
• Sacks argues AI cyber models like Mythos or GPT 5.5 Cyber don't create vulnerabilities but discover existing bugs, and their proliferation will lead to a one-time upgrade cycle as systems are hardened, followed by a new equilibrium between offense and defense.

Science (1)

• David Friedberg cites phase three trial data for Eli Lilly's Retatrutide, a triple agonist peptide, showing an average weight loss of 37 pounds in 40 weeks, an 80% reduction in liver fat, and significant drops in A1C, non-HDL cholesterol, and triglycerides.

Regulation (1)

• Friedberg attended the Supreme Court hearing for Monsanto (Bayer) vs. Hardeman, a case testing federal preemption of pesticide labels under FIFRA against state failure-to-warn laws, with Bayer having paid out $10 billion and reserving another $10 billion for 90,000 outstanding cases.

AI & Tech (5)

• The White House is considering rescinding Anthropic's supply-chain risk designation to allow government use of its models, but some officials oppose a broader rollout of Mythos due to national security and compute capacity concerns.
• Cursor launched an SDK and OpenAI updated Codex for non-developers, asking users to define their role. This signals a battle over interface philosophy: Claude separates technical and non-technical work, while Codex bets on a unified tool for all.
• A viral MS Paint-style prompt instructs AI to redraw images in a 'clumsy, scribbly, and utterly pathetic way,' exemplifying a cultural trend toward low-fidelity, humorous outputs that contrast with the industry's growing maturity.
• OpenAI's Codex model developed an unexplained fixation on mentioning goblins, gremlins, and other creatures. The company traced it to personality reinforcement learning, where a 'nerdy' training preference spilled over, highlighting how quirks can propagate in model-based training.
• A New York Times op-ed predicts a 'permanent underclass' from AI. Nathaniel Whittemore argues Silicon Valley builders often misjudge AI's real-world economic impact, citing economist Kevin Bryan's view that economists largely reject this permanent underclass thesis.

AI & Tech (3)

• Whittemore's AI lab power ranking uses nine categories: compute, enterprise positioning, platform control, consumer positioning, model leverage, momentum, branded narrative, wedge, and X-Factor. Compute and enterprise are weighted highest.
• Amazon launched Quik, an agentic desktop assistant that connects to calendars, email, Slack, and Jira, which The Information framed as a major play for the agentic era.
• Claude announced new integrations with creative and professional tools including Adobe Creative Cloud, Blender, Ableton, and Autodesk.

Models (1)

• DeepSeek released its V4 model family with a 1 trillion parameter Pro version and a smaller Flash version, both featuring a 1 million token context window.

China (1)

• China plans to curb U.S. investment in domestic tech firms like Moonshot and StepFund, following the blocked $2B Meta acquisition of Manus on national security grounds.

Models (1)

• OpenAI released GPT 5.5, seven weeks after GPT 5.4, showcasing a 37-point increase in long-context reasoning and a 60% reduction in hallucinations compared to 5.4.

AI & Tech (10)

• Moonshot AI launched Kimi K2.6, a trillion-parameter open-source model that costs 30 times less than closed models, trained for $4.6 million, demonstrating significant cost-efficiency.
• Dave notes that current AI models can manage dozens or hundreds of other models successfully, enabling consumers to ask a coordinator model to install software or build things without needing technical expertise.
• OpenAI's Chronicle builds memories from periodic screenshots of user activity, raising privacy concerns despite its potential to offer 'telepathy-like' assistance.
• Salim states that 44% of Gen Z workers are sabotaging AI automation efforts by providing incorrect data, highlighting a significant workplace resistance to AI.
• World ID verification is integrating into Zoom to combat deepfake fraud, which caused $130 million in losses between 2019-2023, and is projected to reach $40 billion by 2027.
• Dr. David Lutske's viral post demonstrates Grok creating a realistic AI Frenchwoman with a reflective ID, suggesting video ID verification may soon be unreliable.
• Sheikh Mohammed announced that the UAE will launch an agentic AI government model within two years, with 50% of all government service operations run by AI.
• OpenAI released ChatGPT for clinicians, a free AI co-pilot that outperforms human doctors on health benchmarks, scoring 59 compared to 43.7 for human clinicians.
• Project Top Heart by NYU and Stanford uses AI to analyze 20 variables, aiming to increase the number of viable donor hearts for transplant by an additional 500 per year.
• The FDA-approved blood pressure medication Candesartan has been repurposed using AI to stop and inhibit MRSA infections, which affect 2.8 million people and kill 35,000 annually in the U.S.

Science (2)

• MRNA vaccines for pancreatic cancer show lasting results, with 87.5% of patients who generated a strong immune response still alive after six years, compared to a historical 13% survival rate.
• A single-shot CAR-T infusion achieved 100% cancer-free status in all 20 melanoma patients within two months of treatment, with no recurrence after a median follow-up of 15.3 months.

AI & Tech (5)

• A sophisticated spyware attack in Indonesia used a fake tax app to steal biometric data and drain over $26,000 from a charity accountant's bank accounts.
• Criminal groups now operate a 'malware as a service' model, buying and selling stolen data and malicious software on platforms like Telegram to execute rapid, personalized attacks.
• The global cybercrime industry is estimated to generate $500 billion annually, a scale comparable to the global illicit drug trade.
• Security firm Infoblox identified a software cluster targeting victims in over 20 countries, with criminals integrating AI chatbots and deepfake tools to enhance attacks.
• Allbirds is abandoning its footwear business, selling all shoe assets and rebranding as Newbird AI to pivot towards AI compute infrastructure.

AI & Tech (2)

• AGI attainment will now be independently verified by a third party, removing OpenAI's sole authority to declare its arrival.
• A viral AI thread demonstrates using ChatGPT to analyze palm photos and generate palm prints, raising significant biometric privacy concerns.

Startups (1)

• Benchmark invested $75 million in Manus in May 2025, which facilitated the startup's move from China to Singapore.

China (1)

• Hosts speculate the Chinese government could escalate by pressuring Singapore for extradition or seizing assets, citing historical actions like the disappearance of Alibaba's Jack Ma.

Autonomous Vehicles (2)

• The Epoch Times reported the car accelerated from 24 km/h to 31 km/h before the incident, and that the child was a local shopkeeper's toddler.
• Jason Calacanis argues self-driving cars will be judged by superhuman, not human, reaction times, and a fatal pedestrian incident in the US would end the industry.

VC (3)

• Ben Horowitz says the traditional venture capital model was built for an era where only about 15 companies per year reached $100 million in revenue.
• Horowitz says a16z's first system innovation was sharing economics with partners but centralizing control. He argues shared control makes organizations impossible to change.
• To bootstrap its network, a16z used unconventional tactics. The firm redirected management fees into network-building and gained corporate introductions by leveraging its HP briefing center connection.

Startups (3)

• Horowitz argues that with software's rise, the number of potential $100M-revenue companies would increase dramatically. He and Marc Andreessen believed the figure could be 200, not 15.
• The firm viewed itself as a network to be bootstrapped. Horowitz explains successful network bootstrapping is the hardest part, using the analogy of selling the first telephone with no one to call.
• Horowitz explains their first controversial investment - a quarter of the $300M Fund I into the Skype buyout from eBay - generated a 4x return in 18 months. This validated their approach to LPs.

AI & Tech (3)

• Horowitz says AI has fundamentally changed venture capital. The historical rule that you can't throw money at a software problem to catch up is now false. Capital can accelerate progress.
• He argues that with AI, code and user interface are no longer significant moats. The key questions for startups are now defining new barriers to entry.
• He argues the biggest AI danger is over-regulation and a U.S. failure to compete, which could lead to China achieving superintelligence first, creating a dangerous power imbalance.

Culture (2)

• Horowitz redefines company culture as a specific set of actions and behaviors, not just beliefs or values. He cites the Bushido concept that culture is a set of actions.
• He argues companies are dictatorships, not democracies, because dictatorships are more efficient in competition. Countries need resiliency to bad leadership, but companies must optimize for speed.

Enterprise (1)

• Horowitz has chosen not to pursue leveraged buyouts despite their profitability with AI. He says the LBO culture of firing people to extract efficiency is opposite to venture capital’s growth mindset.