3D7D
 Price:

AI & TECH

SpaceX bets $60B on Cursor's coding data for AI arms race

Wednesday, April 29, 2026 · from 5 podcasts, 7 episodes
5 SOURCESThis Week in StartupsLenny's PodcastAll-InAI Daily BriefNerd Snipe
  • SpaceX secured a right to acquire AI coding startup Cursor for $60B, trading compute for proprietary developer data.
  • China blocked Meta's acquisition of Manis, claiming permanent sovereignty over its startups regardless of headquarters.
  • OpenAI's GPT-5.5 focuses on agentic endurance over hype, targeting professional coders with multi-hour autonomous tasks.

Elon Musk is trading hardware for data. SpaceX, preparing for a $2 trillion IPO, secured the right to acquire AI coding startup Cursor for $60 billion by the end of 2026. If the deal falls through, SpaceX pays a $10 billion collaboration fee. The move, discussed on All-In and This Week in Startups, is a strategic swap: Cursor gets access to SpaceX's Colossus supercomputer to solve its compute costs, while xAI gains the high-quality, real-world coding traces it lacks to build a competitive model.

The acquisition is a defensive play in a global scramble for AI talent and proprietary data pipelines. Just weeks earlier, China's government blocked Meta's attempted acquisition of the AI startup Manis, which had relocated to Singapore. As Jason Calacanis noted on This Week in Startups, Beijing's one-sentence order proves it views any company founded on its soil as permanent state property, ending 'Singapore washing' for Chinese tech founders.

"Beijing’s National Development and Reform Commission issued a one-sentence order to stop the deal. This move effectively ends 'Singapore washing,' the practice of Chinese entrepreneurs moving management to neutral ground to court Western investment."

- Jason Calacanis, This Week in Startups

While nations lock down assets, the model wars intensify. OpenAI released GPT-5.5, which Nathaniel Whittemore on The AI Daily Brief described as a pivot from AGI hype to 'inference company' execution. The model's standout feature is stamina - reliably running autonomous coding tasks for over seven hours, a leap toward persistent agentic work. It outperforms Anthropic's Opus 4.7 on several benchmarks, though lags on Swebench Pro.

The realignment turns legacy software business models into casualties. On All-In, David Sacks and Chamath Palihapitiya detailed a 'SaaS debt bomb,' where AI agents cannibalize per-seat licensing. Thoma Bravo is reportedly handing Medallia back to creditors, wiping out $5.1 billion in equity, as AI-driven internal tools undercut bloated software contracts.

"If a company can spin up internal AI agents to handle customer surveys or HR tasks for a fraction of a license fee, they will. This is the deflationary force that Fed Chair candidate Kevin Warsh recently described."

- David Sacks, All-In

For Snap CEO Evan Spiegel, the lesson that 'software is not a moat' is now a decade old. On Lenny's Podcast, he argued distribution is the only durable advantage, pushing Snap to bet on AR glasses hardware. In AI, the moat is shifting to the fusion of compute, proprietary data, and deep integration - exactly what the SpaceX-Cursor deal aims to lock down.

The $60 billion price tag is a call option on recursive self-improvement. If an AI can master coding, it can rewrite its own architecture. Musk isn't just buying an IDE; he's buying the data to build the model that builds the next model.

AgentsBig TechStartups
Elon MuskNathaniel WhittemoreAnthropicMetaSpaceXxAIChinaCursorGPTOpenAIOpus

Source Intelligence

- Deep dive into what was said in the episodes

This Week in Startups
This Week in Startups

To the left of this of thumbnail i want to add a chinese flag to the left destroying the meta logo. because the title is China Kills Meta / Manus Deal (Story Of The Year) | E2281 • Apr 28

  • Microsoft and OpenAI restructured their partnership, ending Microsoft's exclusive sales rights and allowing OpenAI to pursue deals with Amazon AWS and others.
  • Microsoft's key concession was removing a revenue share it paid to OpenAI. In exchange, Microsoft secured its revenue share from OpenAI until 2030, regardless of AGI progress, and remains the primary cloud provider.
  • Microsoft owns 27% of OpenAI, a stake analysts value in the hundreds of billions if OpenAI achieves multi-trillion dollar status.
  • China's National Development and Reform Commission blocked Meta's acquisition of the AI startup Manus, ordering the deal's cancellation.
  • The Manus deal was partially executed, with capital transferred and some employees already moved to Meta's Singapore HQ before the Chinese government's intervention.
  • Benchmark invested $75 million in Manus in May 2025, which facilitated the startup's move from China to Singapore.
  • The move signals the end of 'Singapore washing,' where Chinese entrepreneurs relocated headquarters to access foreign capital, as China now asserts control over companies formed within its borders.
  • A purported dashcam video from Chongqing on March 9, 2026, shows a Harmony OS self-driving car hitting a toddler, though its authenticity is questioned.
  • 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.
Also from this episode: (3)

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.

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.
StartupsBig TechChinaModels

SpaceX and Cursor team up to topple Claude Code | E2279 • Apr 22

  • SpaceX bets billions on Cursor to secure the data needed for recursive AI self-improvement.
  • Bitstarter launches a crowdfunding model to strip power from predatory Bit Tensor investors.
  • Subnet 11 creates a sandbox for AI agents to write their own instruction sets.
AI InfrastructureModelsStartupsAgents
Lenny's Podcast
Lenny's Podcast

Snapchat CEO: Why distribution has become the most important moat | Evan Spiegel • Apr 26

  • Snapchat boasts over one billion monthly active users and generates more than $6 billion in annual revenue, with users posting over eight billion AR lens photos daily.
  • Building durable social consumer products is challenging because companies overemphasize product-market fit and neglect distribution, which Evan Spiegel identifies as the most critical moat.
  • Evan Spiegel cites TikTok's success, driven by billions in subsidies for creators and viewers, and Threads' growth, which leveraged Meta's existing extensive distribution network.
  • In its early days, Snapchat focused on connecting users with their closest friends, creating value through deep relationships rather than broad network effects to differentiate from larger social networks.
  • Evan Spiegel states that Snapchat learned 15 years ago that software is not a durable moat due to easy cloning, a lesson now being rediscovered with AI advancements.
  • To build durable moats, Snapchat shifted strategy from easily copied software features to developing ecosystems with creators and developers, and investing in difficult-to-replicate hardware like vertically integrated AR.
  • Snapchat's long-term hardware investment, including Specs, aims to create technology that connects people and keeps them grounded in the real world, countering the isolating effects of current mobile devices.
  • Snap initially delayed hiring product managers, believing designers should drive product direction; today, PMs coordinate complex projects, synthesize data, and bring cross-functional teams together.
  • Evan Spiegel advocates deep listening to customers for inspiration, exemplified by Snapchat Stories, which addressed user needs (easy sharing, less pressure) without directly implementing requests like a "send all" button.
  • Evan Spiegel reflects that the CEO role transforms dramatically over time, shifting from hands-on product work to leadership, culture development, strategy, and becoming a "chief explainer" for the company.
  • Evan Spiegel characterizes Snap as the "middle child" in the market, large enough for impact but overshadowed by giants like Meta and Google, necessitating self-definition, particularly through Specs.
  • Evan Spiegel's all-time favorite Snapchat lens is "The Vomiting Rainbow" for the joy it brought; "Face Swap" was an intense early innovation, enabled by Snap's Gen AI lab created 10 years ago.
Also from this episode: (12)

AI & Tech (6)

  • Specs anchor digital content directly in the real world rather than displaying disruptive notifications on a small screen, fostering shared experiences and hands-free interaction as a new computing paradigm.
  • AI empowers designers to ship code and reduces creative friction, with Snap implementing AI tools and guardrails like automated code review to maintain product stability at scale.
  • Snap's AI strategy focuses on defining "jobs to be done" for users and advertisers, then building specialized AI agents to automate workflows, from product ideation to go-to-market execution, using Claude.
  • Evan Spiegel uses a personal AI agent in Glean that combs through internal dashboards, documents, and weekly reports to highlight key focus areas and potential issues.
  • Evan Spiegel holds a contrarian view that humanity is more important than technology, arguing that human comfort and adoption, not just technological advancement, will ultimately dictate AI's deployment and impact.
  • Evan Spiegel recommends David Pogue's "The First 50 Years of Apple" for its historical insights and "The End of the World Is Just the Beginning" for its analysis of global shipping vulnerability.

Enterprise (3)

  • Snap's innovation culture combines a small, flat design team with a large operational organization, fostering dialogue and mutual respect between them, as described by Safi Bahcall's "Loonshots."
  • A core element of Snap's design process is high velocity, with weekly meetings reviewing hundreds of ideas and requiring new designers to present work on day one to foster rapid iteration and reduce preciousness.
  • Evan Spiegel describes the current year as a "crucible moment" for Snap to prove profitability and sustained growth across its platforms, providing a solid foundation for the consumer launch of Specs.

Society (1)

  • Evan Spiegel implements varying screen time policies for his four sons, with zero screen time for his 2-year-old, infrequent use for his 6 and 7-year-olds, and full tech immersion for his 15-year-old.

Media (1)

  • Evan Spiegel enjoyed "Marty Supreme," describing it as an intense, "full throttle movie experience" that kept him on the edge of his seat.

Culture (1)

  • Through his children, Evan Spiegel is rediscovering the "art and personality" of Pokemon, noting its potential for brand and franchise growth.
Big TechStartupsAI InfrastructureMarkets
All-In with Chamath, Jason, Sacks & Friedberg
All-In with Chamath, Jason, Sacks & Friedberg

SpaceX-Cursor Deal, SaaS Debt Bomb, New Apple CEO, SPLC Indictment, Colon Cancer Spike • Apr 24

  • David Sacks, who was in D.C. at the White House, described President Trump as pleasant, genial, and interested in AI issues, contrasting with media portrayals.
  • Sacks noted that President Trump advocates for American AI companies to generate their own power, opposing approaches that halt progress and support DEI values through AI.
  • SpaceX has entered a deal to acquire Cursor, an AI coding startup, by the end of 2026 for $60 billion or pay $10 billion for collaboration, aiming to create the world's best coding AI.
  • Cursor's run rate was $2 billion in February, projected to reach $6 billion by late 2026; this deal could significantly boost SpaceX's projected 2026 revenue of $22-24 billion.
  • Chamath Palihapitiya believes the Cursor deal structure prevents SpaceX's S-1 IPO filing from going stale, effectively giving Elon Musk a 50% discount on the acquisition.
  • David Sacks argues the Cursor acquisition is complimentary, providing XAI with coding expertise, enterprise clients, and training data, while XAI offers compute resources and a foundation model.
  • Chamath Palihapitiya highlighted that much of AI's value is realized in writing software, but enterprises are creating inefficient agents, underscoring the need for strong developer environments like Cursor's IDE.
  • Chamath Palihapitiya suggests that many vertical SaaS companies are struggling because AI agents make it cheaper and easier for enterprises to spin up internal alternatives, crushing sales and increasing attrition.
  • David Sacks advises founders against venture debt, as it reduces maneuverability, imposes business covenants, and makes companies brittle, contrasting with equity sales that align more stakeholders.
  • Tim Cook's 15-year tenure as Apple CEO saw the company's market cap increase over 10x and revenue grow from $100 billion to over $400 billion, driven by improved services mix.
  • Jason Calacanis believes Apple under Tim Cook missed key innovations like more practical AR glasses, a killer AI assistant, a self-driving car, a search engine, a television, and consumer robotics.
  • Chamath Palihapitiya argues that Tim Cook was an excellent steward, significantly shrinking Apple's share count by 44% and investing in R&D and proprietary silicon, but faces the challenge of adapting to a more heterogeneous device future.
Also from this episode: (14)

AI & Tech (4)

  • David Sacks anticipates a race to develop dedicated, cost-effective cyber models comparable to Mythos, as AI-powered hacking risks drive demand from IT departments and CSOs.
  • Kevin Warsh argues that AI's deflationary effect is reducing business costs, leading to economic expansion as companies reinvest savings from SaaS budgets, but also notes that traditional inflation metrics are flawed.
  • David Sacks identifies a challenge for private equity in SaaS, noting that while public SaaS company valuations are attractive (e.g., Salesforce down 32% in six months), predictable cash flows are jeopardized by AI alternatives.
  • Chamath Palihapitiya claims that venture capital and private equity increase SaaS prices to meet return hurdles, making products overpriced and vulnerable to AI-driven cost cutting and unit price reductions.

Business (2)

  • Toma Bravo is reportedly handing Medallia, a customer experience SaaS company acquired for $6.4 billion in 2021, to creditors, wiping out $5.1 billion in equity due to rising debt servicing costs.
  • Chamath Palihapitiya shared his personal experience with a $420 million credit line almost collapsing, reinforcing his belief that debt makes businesses and individuals vulnerable to market disruptions.

Labor (2)

  • David Sacks points out that government pension plans, unlike corporate 401Ks, are underfunded due to public employee unions, threatening to bankrupt U.S. governments.
  • Jason Calacanis suggests that government waste, fraud, and abuse in California, exemplified by the homeless industrial complex, could be addressed by eliminating a minimum of 20-30% of inefficiencies.

Corruption (3)

  • The Southern Poverty Law Center (SPLC) is facing allegations of wire fraud and money laundering between 2014 and 2023, specifically for funneling over $3 million to informants in hate groups.
  • SPLC allegedly paid an informant, F-37, over $270,000 between 2015 and 2023, who was a member of the online leadership chat group that planned the 2017 Unite the Right event in Charlottesville.
  • David Sacks states the SPLC's fundraising doubled to $136 million after Charlottesville from $58 million in 2016, suggesting the alleged actions were a 'grift' to increase donations.

Politics (3)

  • Chamath Palihapitiya calls for the dismantling of NGOs that 'cosplay as overlords' and urges donors to sue the SPLC, citing $822 million allegedly held in offshore bank accounts.
  • David Friedberg criticizes 501(c)(3) non-profit organizations for straying from their IRS-defined charitable activities, suggesting many operate with commercial or misaligned interests.
  • David Sacks posits that civil rights organizations, once achieving their goals, shifted from ensuring equality of opportunity to demanding equality of outcomes, rebranded as 'anti-racism'.
AgentsBig TechMacroStartupsReasoning
The AI Daily Brief: Artificial Intelligence News and Analysis
The AI Daily Brief: Artificial Intelligence News and Analysis

Nathaniel Whittemore

What I Learned Testing GPT-5.5 • Apr 24

  • OpenAI released GPT 5.5 on Friday at 2 p.m., describing it as a 'new class of intelligence for real work' empowering agents to understand complex goals and use tools for task completion.
  • GPT 5.5 significantly outperformed Anthropic's Opus 4.7 on several agentic coding benchmarks, including Terminal Bench 2.0 and GDP Val.
  • Artificial Analysis ranks GPT 5.5 as the clear number one model on its intelligence index, breaking a three-way tie with Anthropic and Google by three points.
  • Many users found GPT 5.5 to be the new standard, significantly faster and easier to collaborate with than Opus 4.7, and the strongest model for engineering tasks.
  • Matt Schumer notes that while GPT 5.5 is a 'massive leap forward,' 99% of users may not notice a dramatic difference because previous models were already highly capable for most routine tasks.
  • Bindu Reddy and Code Rabbit found GPT 5.5 superior for coding tasks, with Code Rabbit reporting a 79.2% expected issue found rate in code review, versus a 58.3% baseline.
  • Peter Gsta and Adah Mclofflin observed GPT 5.5's greatly improved reliability on long-running tasks, with tasks successfully running for 7-8 hours or even 31 hours continuously.
  • Nathaniel Whittemore recommends users invest time in Codeex, OpenAI's core workspace, noting its improved context compaction for ongoing, single-thread conversations.
Also from this episode: (8)

Models (7)

  • Despite strong overall performance, GPT 5.5 lagged behind Opus 4.7 on Val's AI's professional task benchmarks and Swebench Pro, a coding benchmark.
  • Theo notes GPT 5.5's cost per million tokens is double GPT 5.4 and 20% higher than Opus 4.7, at $5 in and $30 out respectively.
  • OpenAI's Gnome Brown argues model intelligence should be measured by 'intelligence per token or per dollar' rather than a single number, especially for products like Codeex.
  • Scaling01 estimates GPT 5.5's parameters are 2-5 trillion, compared to Mythos at approximately 10 trillion and GPT 5.4 at 1-2 trillion.
  • Nathaniel Whittemore found GPT 5.5 significantly better at writing, following instructions for a clear, journalistic style without the 'dramatic flare' often seen in Opus models.
  • OpenAI's communication strategy for GPT 5.5 emphasized iterative deployment and democratization, contrasting Anthropic's approach of announcing powerful models without broad public access.
  • GPT 5.5 demonstrated strong data analysis and spreadsheet capabilities for Nathaniel Whittemore, generating insightful podcast strategy recommendations from diverse data and organizing information into spreadsheets.

AI & Tech (1)

  • OpenAI chief scientist Jacob Pachi and President Greg Brockman indicate that GPT 5.5 is a 'beginning point' and forecast 'rapid continued progress' and 'extremely significant improvements' in AI capabilities in the short to medium term.
ModelsAgentsBig TechStartups

What GPT Images 2 Unlocks • Apr 22

  • SpaceX partnered with Cursor, an AI coding tool, acquiring rights to purchase Cursor for $60 billion later this year; if the acquisition fails, SpaceX will pay Cursor $10 billion for their collaborative work.
  • The SpaceX-Cursor deal potentially solves Cursor's reported issue of losing money on every Claude and OpenAI token served, giving them access to XAI's Colossus training supercomputer with millions of H100 equivalent units for in-house model development.
  • XAI could benefit from Cursor by gaining a significant data pipeline to improve its models, especially since XAI has struggled to generate revenue or release impactful models, and lacks a footprint in the AI coding space.
  • SpaceX's IPO disclosure documents reveal Elon Musk increased his stake by $1.4 billion and could receive a compensation package tied to market cap achievements ranging from $1.1 trillion to $6.6 trillion.
  • An unauthorized group accessed Anthropic's Claude Mythos preview via a third-party vendor and information from the Merkle data breach, despite Anthropic's tight control measures for cybersecurity purposes.
  • Sam Altman criticized Anthropic's promotion of Mythos, suggesting its fear-based marketing positions AI control as a justifiable purchase, rather than focusing on legitimate safety concerns.
  • Google released an upgrade to its Deep Research agents, now featuring MCP support for third-party data and the ability to output charts and infographics using Nano Banana models, with a Max version outperforming GPT 5.4 and Opus 4.6.
  • The improvements in Google's Deep Research agents, despite still using Gemini 3.1 Pro under the hood, stem entirely from harness upgrades and additional inference, not a more advanced base model.
  • OpenAI's new ChatGPT Images 2.0 model leads the Arena Elo score human preference board with a record-breaking 242-point lead over the previous leader, indicating a significant jump in quality.
  • Nathaniel Whittemore argues ChatGPT Images 2.0 is the first image model for the 'agentic era' because its primary impact will come from integration with other systems, rather than standalone viral moments.
Also from this episode: (3)

Models (3)

  • GPT Images 2.0 offers enhanced precision and control, handling small text, UI elements, and dense compositions at resolutions up to 2K, along with multilingual capabilities for designs where language is integrated.
  • Users are already integrating GPT Images 2.0 with Codex, creating a pipeline to generate UI mockups and then convert them into working code, addressing Codex's previous limitations in UI design.
  • While GPT Images 2.0 shows vast improvements, Boyan Tongues noted visual artifacts, and Sharon Goldman's sister found anatomical inaccuracies in medical images, highlighting a zero-tolerance for errors in certain use cases.
Big TechStartupsModelsEnterpriseSafety
Nerd Snipe with Theo and Ben
Nerd Snipe with Theo and Ben

We've been testing GPT-5.5 for a few weeks now... • Apr 23

  • SpaceX AI and Cursor announced a partnership, giving SpaceX the right to acquire Cursor later in the year for $60 billion or to pay $10 billion for their joint work.
Also from this episode: (11)

AI & Tech (9)

  • Theo and Ben discuss the Vercel hack, which started with a compromised Google OAuth credential from a deprecated service called context.ai.
  • They note the hack was not a direct compromise of Vercel, but Vercel handled the communications well. They warn other services using context.ai are likely compromised.
  • Theo states AI tools are accelerating hacking attacks, marking the start of a chaotic period for cybersecurity rather than its end.
  • Qwen/Qwen2.5 27B is a new 27-billion-parameter open-weight model that shows flagship-level coding ability, making powerful local inference more feasible.
  • Kimmy/Kimi K2.6 is a 1.1-trillion-parameter Chinese open-weight model that impressed Theo with its coding output quality and feel, rivaling Claude and GPT in initial tests.
  • The hosts frame the SpaceX-Cursor deal as a strategic fit: Cursor needs massive compute and data, while SpaceX AI has unused compute and needs a strong AI product.
  • GPT-4o/Image 2 significantly improves on predecessor weaknesses like text rendering, instruction following, and multi-image character coherence, becoming a versatile multimodal tool.
  • Ben criticizes Anthropic for testing a pricing change that removed Claude Code from the $20 Claude Pro tier for some new signups, calling the handling incompetent or dishonest.
  • Ben is highly positive on the new GPT model (likely GPT-4.5), praising its ability to write concise, high-quality code on low-reasoning mode and its effectiveness for building complex projects.

Models (2)

  • Theo is frustrated with the new GPT model, citing problems with context pollution, over-editing, difficulty correcting course mid-thread, and a heightened fear of context window compaction.
  • Ben says the GPT model (Pro version) demonstrated exceptional problem-solving ability, solving previously unsolved DefCon cryptography challenges and a custom crypto puzzle in under three hours.
StartupsCodingBig TechOpen SourceModels