🎧 Listen to this Newsletter on the Sun app
Here's what's been trending:
Trump flew Jensen Huang, Tim Cook, and Elon Musk to Beijing: The US cleared 10 Chinese firms for H200 chips. Not a single chip has shipped yet.
Claude Mythos became the first AI model to complete both of AISI's offensive cyber ranges: Capabilities are doubling every 4.7 months.
Cerebras priced its IPO at $185/share, raising $5.55B at a $56.4B valuation: The deal was 20x oversubscribed.
Cisco posted record quarterly revenue of $15.84B and raised its AI infra order forecast to $9B: Foxconn reported $67B in Q1 revenue, up 29%.
Musk v. Altman closing arguments are in: Altman holds $2B+ in stakes at companies that did business with OpenAI.
🎧 Listen to this Newsletter on the Sun app
Startup Intros has a simple mission: to help tech founders navigate everything in the early-stage, from compliance to capital.
🎁 Soft Launch: Incorporation & Accelerator Resources
Thanks for showing up!
To everyone who joined our A16Z Speedrun AMAs over the last two weeks. Thank you! Sharp questions that lead to action!
Here's what we're putting in front of you next.
1. Incorporate Your Delaware C-Corp:
Cert of Incorporation, EIN, founder stock, 83(b)
Bylaws, IP assignments, board consents
Clean cap table on day one, ready for your first SAFE
2. The Accelerator Tracker (60+ programs):
Antler Japan: $150K SAFE · May 11
Neo 2026: $750K for 1% · May 15
AngelPad: $120K for 7% · May 30
Boost VC: $500K for 15% · Jun 1
Techstars AI Healthcare Baltimore: $220K · Jun 10
🔥 Startup Intros Events Coming Up
May 15: Caffeine & Capital: Our morning coffee meetup is back at Corgi Cafe in FiDi! Grab coffee, meet founders and investors & register on Luma!
May 19: Agentic AI Founders’ Night-Out: We are going beyond copilots and talking about the rise of Agent Orchestration! Register on Luma!
🌏 The China Tech Reset
✈️ Air Force One Tech Diplomacy
Xi told the room that China's door will "open wider," and Trump made sure Jensen, Elon, and Tim Cook were in that room. Jensen called the summit "one of the most important in human history." Xi said US companies have "benefited from China's reform and opening up" and would enjoy "even broader prospects." The White House echoed back: both sides discussed expanding market access for US businesses while increasing Chinese investment in the US.
Each CEO came with different asks:
Jensen, being a bargaining chip, wants to sell more chips as Nvidia's China revenue share collapsed from 26% to ~5% since the export bans,
Elon came for FSD approval and permission to ship Chinese training data offshore for Tesla's autonomy stack,
Cook was managing damage control on manufacturing concessions, having already shifted U.S.-bound iPhone production to India.
Also, Dina Powell McCormick, Meta's new global emissary, joined the delegation and met King Charles III on the way.
🚧 Not a Single Chip Has Shipped
Clearances have been granted, but deliveries are at zero. The US gave Nvidia licenses, but Beijing has halted actual deliveries, leaving the deal in limbo. "The ball is in China's court" after Chinese orders fell apart post-GTC. The H200 is Nvidia's second-most-powerful chip; the H100 ban still holds.
The Irony: China was already building around the chip ban. Chinese companies are aggressively ramping homegrown AI chips even as Nvidia tries to re-enter the market. Huawei, Alibaba's Tongyi lab, and a dozen stealth-stage startups are all shipping alternatives. The H200 opening may have arrived too late to matter for the most ambitious players.
🧠 The Brain Drain in Reverse
Also, the number of Chinese graduates returning home from abroad rose 12% YoY in 2025, fueled by big government incentives and a cooling US research funding environment. Beijing is offering "sea turtle" researchers six-figure signing packages, housing subsidies, and fast-tracked lab access. The US, meanwhile, has been defunding research programs and cutting visa approvals.
Translation: China is not just waiting for Nvidia; it’s building the talent pipeline to compete on the frontier. In a world where China has both the talent and eventually the chips, what does US AI leadership actually rest on?
⚠️ AI Is Getting Dangerous, Fast
🎯 Claude Mythos Clears Both Cyber Ranges
Claude Mythos Preview is the first AI model ever to complete both of AISI's cyber ranges, the UK government's benchmarks for measuring AI offensive capability. The ranges simulate real-world cyberattacks. GPT-5.5 completed only one. A newer Mythos checkpoint completed a 32-step corporate network attack, estimated to take a human expert 20 hours, in 6 out of 10 attempts.
The DC Turf War: The White House's Office of the National Cyber Director and the Commerce Department's CAISI are actively fighting over which agency should lead AI model evaluations. Unfortunately, the US has no equivalent body to Britain's AISI.
The Context: XBOW independently evaluated Mythos and found it "substantially better than prior models at finding vulnerabilities." Anthropic has put the capabilities behind Project Glasswing and is working with defenders. But the lab's own model is now the most capable autonomous offensive hacking tool ever independently benchmarked.
⏱️ The 4.7-Month Doubling Clock
AISI estimated in February 2026 that the length of cyber tasks AI can complete has been doubling every 4.7 months since late 2024, down from an 8-month doubling time in November 2025. Then Mythos and GPT-5.5 exceeded even that accelerated timeline. AISI's candid note: "It is currently unclear whether this is a new trend or a one-off."
Matt Clifford's read on X: "There is no deceleration." Ethan Mollick flagged that both models appear to be token-limited rather than capability-limited; throw more compute at them, and they go further. The ceiling isn't in sight yet.
Translation: AISI released a report that should be uncomfortable reading at every major AI lab, every cybersecurity team, and every government agency pretending to have this under control.
⚖️ Musk v. Altman: The Verdict
📋 $2B in Conflicts, $100B in Partnership
Court filings revealed Sam Altman holds $2B+ in stakes in companies that did business with OpenAI, including a $1.7B position in fusion startup Helion Energy, which signed a power purchase agreement with OpenAI. Musk's legal team is using this to argue that Altman steered OpenAI's resources toward companies in which he had personal financial interests.
Microsoft executive Michael Wetter testified that Microsoft has spent $100B+ on OpenAI, combining equity investments and infrastructure buildout. Separately, reporting confirmed OpenAI and Microsoft have capped their revenue-sharing arrangement at $38B, clearing the path for OpenAI's full for-profit conversion. Microsoft also feared being too dependent on OpenAI, per internal emails surfaced at trial.
🎭 The Real Stakes & Ruling
Altman's team argued Musk abandoned OpenAI after losing a control dispute and is now trying to interfere with its business because xAI is a direct competitor. OpenAI's lead researcher testified that at a 2017 meeting, Musk pushed to "race toward AGI to beat DeepMind," while safety staff objected. The picture: two different visions for the same nonprofit, and a bitter split when Musk lost.
The Judge Decides: No jury. The relief Musk is seeking includes blocking OpenAI's for-profit conversion, unwinding Microsoft's influence, and a raft of financial damages.
A ruling for Musk: OpenAI's cap table gets messy; the nonprofit conversion stalls; the $150B valuation is at risk while investors wait for clarity.
A ruling for Altman: OpenAI moves full speed into Deployment Company mode; Microsoft's $100B bet pays off; xAI is positioned as the scrappy underdog fighting the monopolist.
Translation: This is not a dispute about the legal technicalities of a nonprofit conversion. It is a dispute over who gets to write the origin story of the decade's most important tech company, and whether that story ends with Altman as a self-dealing founder or a visionary who made hard calls under pressure.
💰 The AI Money Machine
🚀 The AI IPO Wave
Cerebras priced at $185/share, raising $5.55B at a $56.4B fully diluted valuation, above its expected range of $150-$160. The deal was 20x oversubscribed, making it the largest US tech IPO since Snowflake. The catch: 62% of revenue comes from a single UAE government customer. One X post from a sharp analyst noted: "Public AI exposure is so scarce relative to demand that the market will pay through the concentration risk."
Blackstone Digital Infrastructure Trust raised $1.75B in its US IPO, selling 87.5M shares at $20 each and planning to acquire data centers in the $250M to $1.5B range. The picks-and-shovels AI trade is going public in bulk this week.
📡 Infrastructure Earnings
Cisco posted record Q3 revenue of $15.84B, up 12% YoY, and raised its FY2026 AI infrastructure order forecast from $5B to $9B. Stock jumped 15%+. The company cut nearly 4,000 jobs to fund the AI pivot, and hyperscaler networking orders grew 50% YoY. Cisco becomes one of the largest AI infrastructure vendors by order volume.
Foxconn reported Q1 revenue of $67B, up 29% YoY, with AI servers now its largest revenue segment. CEO Michael Chiang said Q2 will show "significant growth vs Q1 on robust AI demand." The company that makes your iPhone is now primarily in the business of making AI compute.
Jensen Huang's charitable foundation also made headlines: it purchased $108.3M in CoreWeave compute time and donated it to universities. A strategic philanthropic move that puts CoreWeave hardware in front of the next generation of AI researchers.
Translation: Cerebras priced at $56B on a book 20x oversubscribed. The money is not hedging. It is betting. The concentration risk in that Cerebras deal is a tell: there is still far more capital chasing AI exposure than there are clean, diversified AI businesses to invest in.
🚀 Product Launch Quick Hits
Anthropic launches Claude Agent SDK credits: Paid plans get a monthly credit ($20-$200) for Agent SDK and third-party tools like OpenClaw starting June 15; power users say it's a 25x cut in subsidized usage.
OpenEvidence hits 65% of US physicians: The AI-powered medical search engine now reaches roughly 650,000 US doctors, plus 1.2M internationally, and most patients don't know their doctor uses it.
Microsoft Edge now reads all your open tabs: New Copilot features let Edge gather context from every open tab, generate quizzes, and offer podcast summaries; standalone "Copilot Mode" is gone because everything is Copilot Mode now.
Google to announce new Gemini at I/O next week: Sources tell The Verge the model lands roughly in GPT-5.5's class, with AI agent Gemini Spark also prepped for launch.
TikTok GO launches US travel booking: TikTok partnered with Booking.com and Expedia to let users book hotels, attractions, and experiences directly in the app.
OpenAI backs an IAEA-style AI governance body: OpenAI wants a US-led global AI governance organization, including China; the proposal landed the same week Altman's personal conflicts hit the public record.
AT&T, T-Mobile, and Verizon form a satellite joint venture: The three carriers agreed to pool spectrum and satellite capacity to eliminate US wireless dead zones, with technical details still being worked out.
Instagram launches Instants: Meta's new ephemeral photo-sharing feature lets users send disappearing photos to close friends, available as both an in-app feature and a standalone iOS/Android app.
👔 Personnel Quick Hits
Gallup: 70% of Americans oppose data centers near their homes: A new survey found opposition now higher than for nuclear power plants, with nearly half strongly opposing local construction, citing water, electricity, and environmental concerns.
Meta's May 20 layoffs are hitting differently: Wired spoke with 16 current and former employees ahead of the 10% cuts; the most common word was "miserable," and anyone with options is hoping to be laid off for the severance.
LinkedIn cuts 5% under new CEO: Dan Shapero, three weeks into the job, announced roughly 875 layoffs across engineering, product, and marketing.
Cisco cuts 4,000 as stock hits record: CEO Chuck Robbins announced layoffs alongside $15.84B in record revenue, directing the savings toward AI, security, and silicon.
🌟 Editor's Note
At Startup Intros, our mission is to bring the latest founder-investor news straight to your inbox, keeping you ahead in the fast-paced world of Silicon Valley.
🎧 Listen to this Newsletter on the Sun app
🚀 Thanks for joining the A16Z Speedrun AMAs.
Two things for the founders who showed up:
→ Incorporate your Delaware C-Corp: And get your EIN, founder stock, 83(b); all in one workspace. Start →
→ Accelerator Tracker: 19 programs closing in the next 90 days, sorted by deadline. See the list →
From day zero to term sheet.
💭 Parting Thoughts
What a week: H200 clearances for China, but no chips shipped. Cerebras IPO at $56B, 20x oversubscribed. Claude Mythos cleared both cyber ranges while the US government argues about which agency is in charge. Cisco called $9B in AI orders and cut 4,000 people on the same afternoon.
The founders I talk to are watching two things:
The China situation, because if those H200 chips start shipping, compute pricing moves. If they don't, the US AI supply chain stays insulated, for now.
The Musk v. Altman ruling is a win for Musk because it doesn't just hurt OpenAI; it sets a precedent for how founder-led companies can be challenged on governance grounds at any stage.
The big surprise this week wasn't the AI safety findings or the IPO. It was Claude Mythos quietly becoming the most capable offensive cyber tool ever independently benchmarked, while a DC turf war over who evaluates it plays out in the background.
Till next time!
 | Dev Chandra |
Startup Intros Trending Thursday: Your trusted source for founder-investor insights, delivered with clarity and focus.