Nvidia and Google bet on open AI to break OpenAI's grip

Other (20)

• Marty argues that Bitcoin serves as the ultimate victor and safe haven in an economic environment where central banks actively devalue their national currencies.
• Marty and Matt discuss a fund manager named Leopold, dubbed "the boy wonder," who reportedly lost over $40 billion, possibly due to excessive leverage and a margin hunt by Wall Street firms.
• Current Bitcoin network stats show a price of $64,760, a market cap of $1.3 trillion, and an average block time of 10 minutes and 16 seconds. The mempool currently has 86,317 transactions.
• Marty clarifies that Arthur Hayes's family office, Maelstrom, funds a Bitcoin grant program run by Jonathan Beer, while Farsight Insights is Beer's separate investment management company.
• BIP 110 activation is nine days away, with a monitoring dashboard (bip110.orange.surf) indicating a high likelihood of a chain split where BIP 110 nodes would form a minority chain.
• Strike has launched beneficiary planning for Bitcoin inheritance, a feature addressing the need for users with significant collateral balances to designate an heir for their funds.
• The Clarity Act now includes a provision safeguarding self-custody Bitcoin from abandoned property laws due to inactivity, establishing federal preemption over state laws.
• Secretary Scott Bessent strongly supports the Blockchain Regulatory Clarity Act, arguing it enhances compliance for digital asset intermediaries while codifying existing Treasury policy for non-custodial developers.
• Ventium, a Bitcoin accelerator based in Brazil, announced new fellowship cycles, selecting 12 developers from 322 applications for its second cohort and funding two new grantees.
• NVIDIA CEO Jensen Huang, alongside Elon Musk, Microsoft, and Google, publicly supported open-weight, open-source AI models, highlighting a powerful industry alignment toward freedom in AI development.
• Marty and Matt detail how OpenAI and Anthropic are reportedly "cannibalizing" customers by using their proprietary data to develop competing products, citing examples like Figma, Novo Nordisk, Microsoft, and Harvey.
• Nigerian President Bola Tinubu established a Virtual Asset Council, chaired by the Central Bank, to regulate digital assets, which Marty views as an attempt to implement a surveillance and control regime.
• Sparrow Wallet version 2.5.3 was released with Aera hardware wallet support, XDG directories, and PSBT verification; Marty warns users about scam mobile applications.
• Buzz, a Noster-based application, leverages cryptographically signed events to build private "company brains" and enables a "multiplayer harness" for agents to interact and share compute resources within trusted environments.
• India requested GitHub to ban BitChat, prompting developer Callie to enable offline self-transfer of the Android APK via Bluetooth and develop a full-fledged client for smartwatches to circumvent censorship.
• An open-source ESP32 LoRa mesh relay was introduced to bridge with BitChat, allowing users to extend the application's range through low-power, continuously running devices.
• Russia charged Telegram founder Pavel Durov with facilitating terrorism after he refused demands for mass surveillance and censorship, a move Durov publicly defied.
• Colorado's new law, effective August 1st, mandates a restrictive process for purchasing semi-automatic firearms with detachable magazines, requiring multiple background checks, fees, classes, and an exam.
• Marty highlights an alleged coordinated cyber attack on over 30 Minnesota water systems, with speculation of Iranian links, underscoring the vulnerability of critical infrastructure.
• Marty advocates for unleashing open-weight AI models to audit and defend critical infrastructure, suggesting that a short-term rough patch would ultimately lead to stronger systems.

Models (8)

• Nathaniel Whittemore reports OpenAI's GPT 5.6 family (Soul, Terra, Luna) will launch Thursday, with early testers providing positive feedback. Ali K. Miller calls it an "execution beast" and a significant leap over previous models.
• Pietro Schirano describes GPT 5.6 as the "best model I've ever used," praising its speed, creativity, and front-end design fixes. Ethan Mollick notes Fable is more independent, while 5.6 Soul is faster and works interactively.
• Elon Musk confirmed Grok 4.5's public release, stating it's based on a 1.5 trillion parameter V9 foundation model with Cursor data. It is an "open class model" that is faster, more token efficient, and lower cost, performing close to or exceeding Opus.
• Anthropic extended access to Fable 5 on paid plans through July 12th, originally set to end earlier. Andrew Curran suggests this extension fosters a "heroic aura" after many users had already maxed out their usage.
• Meta launched Muse image, its first image model since restructuring the AI division, ranking second on Arena AI's image edit benchmark behind GPT image 2. Alexander Wang explains it is paired with Muse Spark LLM for reasoning.
• Meta's Muse image allows self-refinement, multi-reference composition, and multi-turn editing. It integrates into Instagram and WhatsApp, but its ability to tag others in prompts using public photos for generation raises deepfake concerns.
• Nathaniel Whittemore reports rumors that China's MiniMax is developing M3 Pro, a 2.7 trillion parameter LLM, potentially releasing in Q3. Although MiniMax plans to open source it, the Chinese government's evolving stance on model distribution could alter this.
• Microsoft's "Frontier Tuning" allows companies to customize MAI models, turning generalists into tailored partners. Mustafa Suleyman reports an MAI-tuned model for Excel is 10x more efficient than GPT-5.4, achieving similar benchmarks.

Markets (1)

• Nathaniel Whittemore reports that after its post-IPO quiet period, SpaceX AI received bullish analyst ratings. Morgan Stanley set a $300 target, Bernstein a $239 target, and JP Morgan projected 5,000 Starship launches by 2031.

China (3)

• Reuters reported Beijing is exploring blocking overseas distribution of leading Chinese AI models, considering limits on investments and criminalizing technology leaking. China now views frontier AI as a national security asset, not just a consumer product.
• Rui Ma summarized a Chinese court dialogue on AI open source, highlighting concerns about "open source washing" and China's ambition to develop its own legal framework. Ethan Mollick anticipates the flow of frontier open-weight models may not continue indefinitely.
• If China restricts access to cheaper open-source models, Nathaniel Whittemore argues it would significantly boost US and Western open-weight and alternative model development. This strategic shift addresses high AI costs and compute shortages.

AI Infrastructure (1)

• Model routers are gaining importance for efficiently selecting the right model for tasks, saving costs. Nathaniel Whittemore suggests they could also play a governance role by selecting models based on risk, indicating growing complexity for enterprise AI buyers.

Open Source (8)

• Ben Horowitz argues that open-source models are critical for AI safety and preventing a single AI monopoly, emphasizing the need for broad AI access. He cites historical precedents like the internet and Linux being safer than Windows.
• NVIDIA released an open letter, "OpenWates in American AI leadership," advocating for open-source AI, which was co-signed by Andreessen Horowitz and other prominent companies.
• Ben Horowitz considers open-source AI the single most important policy issue, framing the debate as an anti-competitive stance, often cloaked in safety concerns, versus promoting open models.
• Ben Horowitz argues that without open-source AI, academia is completely excluded from significant participation and innovation in the AI field.
• Ben Horowitz states that banning open-source technology, similar to past attempts to ban cryptography, is impossible because the underlying principles are public, though restrictions could prevent legitimate use.
• Ben Horowitz refutes claims that open-source AI poses unique national security risks from China, arguing such threats are general. He asserts an AI monopoly is a far greater risk, citing Anthropic's refusal to cooperate with the U.S. government.
• Ben Horowitz explains that Chinese labs have outpaced American counterparts in open-source AI performance because U.S. labs built proprietary models, necessitating high valuations to acquire expensive GPUs, and historically, market leaders rarely adopt open-source strategies.
• Ben Horowitz states open source AI is vital for application developers, as proprietary model providers like Anthropic could replicate Microsoft's old playbook by entering application categories and subsidizing their own offerings, while overcharging rivals.

Markets (1)

• Ben Horowitz estimates that the current AI market is less than 3% penetrated, indicating substantial growth potential for all participants without immediate market share conflicts.

Safety (2)

• Ben Horowitz highlights a security incident where OpenAI hacked Hugging Face; proprietary models' guardrails blocked security measures, forcing Hugging Face to rely on an open-source model for defense.
• Ben Horowitz contends that proprietary models from companies like Anthropic and OpenAI have not solved critical AI safety issues such as "reward hacking," suggesting open source allows broader community inspection to find solutions.

Regulation (1)

• Ben Horowitz identifies a significant, well-funded effort, particularly from Anthropic, to ban open-source AI, comparing it to historical attempts by intelligence communities to control cryptography.

Models (2)

• Ben Horowitz agrees with Aaron Levy that distillation, or training on AI model outputs, is generally legal because AI outputs lack copyright protection. He notes Anthropic paid $1.5 billion in a class-action lawsuit for using copyrighted books in training data.
• Ben Horowitz warns that monopolistic AI futures, where a single company automates its AI research and rapidly gains capabilities, threaten a pluralistic startup ecosystem; open models offer the best defense.

Open Source (2)

• Chinese company Moonshot AI released Kimi K3, an open-source model offering performance comparable to Opus 4.8 and GPT 5.6 at a 50% lower cost, sparking debate in the US.
• Friedberg argues open-source AI is crucial for global economic growth, diffusing value to a broad range of enterprises and preventing wealth concentration, echoing the proliferation of the open internet.

Regulation (4)

• The White House is considering banning Chinese open-source AI models due to concerns about intellectual property distillation, though Howard Lutnick advocates for incentivizing US frontier labs instead.
• Palihapitiya argues that closed frontier AI labs are attempting 'valuation preservation' through regulatory capture, as foundational models are commoditizing rapidly with value shifting to application and infrastructure layers.
• Palihapitiya warns that a US government ban on open-source AI would tank the stock market, forcing American enterprises to pay an 'AI token tax' due to higher-cost, proprietary alternatives.
• Palihapitiya contends that increasing housing supply through aggressive permitting reform is the only effective way to lower rents, citing Austin's success and comparing it to New York's restrictive policies.

Safety (1)

• David Sacks warns that rhetoric from companies like Anthropic, focusing on AI guardrails, aims to justify a future ban on open-source models, harming America's AI competitiveness.

Models (1)

• Chamath Palihapitiya explains 'distillation' as training one's own model by observing and collecting output from another; he states that effective prevention would require implementing KYC to slow growth.