Anthropic 完成史上最大规模 IPO,融资 $81.3B
Anthropic pulls off the biggest IPO ever | The Rundown AI
Anthropic 在 Nasdaq 以 ANT 为代码上市,IPO 融资 $81.3B,创下文中所称的最高纪录;股票首日收于 $97.17,较发行价上涨 17%,公司估值超过 $1T。CEO Dario Amodei 表示,公司将依据模型安全状况控制发展节奏,不会向商业或政治压力让步。
推荐理由
精选AI 评分 92
上市首日的融资额、估值与员工持股数据,呈现了资本市场对前沿 AI 公司的定价及其安全承诺面临的商业压力。
主题
正文
AIAnthropic pulls off the biggest IPO ever
PLUS: Use AI to find a job and prepare to get hired
Zach Mink
Good morning, AI enthusiasts, and welcome to our 6,136 new readers. Frontier AI is now public. Anthropic just listed on the Nasdaq in the largest-ever IPO, topping SpaceX’s record from just five months ago.
The debut minted thousands of millionaires overnight and came with a promise from CEO Dario Amodei that safety, not business or political pressure, will set the pace of its models and way forward.
In today’s AI rundown:
Anthropic raises $81B in the largest IPO ever
Rowan’s Corner: 4 things that stuck with me from DealBook
Use AI to find a job and prepare to get hired
OpenAI’s agents now handle half its long research tasks
LATEST DEVELOPMENTSANTHROPIC
🤑 Anthropic raises $81B in the largest IPO ever
Image source: Nasdaq / YouTube
The Rundown: Anthropic just started trading on the Nasdaq under the ticker ANT after raising $81.3B in the largest IPO on record, with shares closing their first day up 17% at $97.17 and valuing the company behind the Claude AI assistant at well over $1T.
The details:
Demand was so strong that Anthropic priced shares at $83, above its target range, and the stock opened at $111.50, briefly valuing the company near $2T.
96% of Anthropic’s 5K+ employees are now millionaires on paper, with each co-founder worth ~$20B; they’ve pledged 80% of their stock to philanthropic causes.
At the IPO price, Amazon’s and Google’s early stakes hit $161B and $147B, and Sound Ventures, co-founded by Ashton Kutcher, turned a $435M bet into ~$7B.
Before the opening bell, Amodei said AI could speed up cures for disease but pledged to pace development until its models are safe, saying Anthropic “will not bend to either business nor political pressure.”
Why it matters: Two months ago, Sam Altman ruled out an OpenAI IPO this year, calling it an “ill-advised moment” given safety concerns. Anthropic went ahead, and the lab built on safety will now face Wall Street’s scrutiny. Every earnings call will be a test of both the AI boom and Anthropic’s pledge to slow down if safety demands it. TOGETHER WITH GLEAN
📊 Turn company data into decision-ready answersVisit glean.com
The Rundown: AI answers are only useful when they’re grounded in trusted data and enterprise context. This eBook shows how Glean brings context to the warehouse and business intelligence tools you already use, helping teams scale self-service analytics without rebuilding their infrastructure.
Read the eBook to:
Understand why AI analytics needs enterprise context
See how Glean works with your existing data and business intelligence tools
Learn how to scale secure, self-service analytics across your business
Download the eBook. ROWAN’S CORNER
🗽 4 things that stuck with me from DealBook
Rowan’s POV at DealBook
Rowan: I spent Wednesday at the NYT DealBook Summit, in a room that was a good mix of tech reporters, top AI execs, and Wall Street heavyweights.
Four things stuck with me:
When Christine Lagarde was asked if we’re in an AI bubble, the room laughed nervously. I don’t think we are, at least not the dot-com kind. Back then, companies built capacity nobody was using yet… today, the labs can’t build compute fast enough to keep up with demand. Still, a room full of top execs laughing nervously is hard to ignore.
Almost every speaker talked about Jensen leaving Nvidia. Musk compared it to “kicking Michael Jordan off your team when you’re on a winning streak.” It’s a reminder that even at trillion-dollar scale, the AI industry still runs on a handful of people.
I’ve never heard Sam Altman push as hard for humans to stay in control. He called gradually handing decisions over to AI “a terrible idea” and said he doesn’t want to end up as AI’s Labrador, or his kids either. A clear shot at Elon Musk, who has said humans could end up “like a pet Labrador” to AI.
Greg Brockman said the biggest differentiator between the labs will be who owns the compute. To me, that means the AI race is becoming less about who has the smartest researchers and more about who has the most chips and power. AI TRAINING
💼 Use AI to find a job and prepare to get hired Visit The Rundown University
The Rundown: In this guide, you’ll use AI to find real job openings, compare them with your experience, tailor your resume, and practice for an interview. One fictional applicant carries through the full process, so you can follow along before using your own details.
Step-by-step:
Choose one target job title, two related titles, and your limits for location, work setup, salary, and experience
Ask ChatGPT to find five current openings. Check each role on the employer’s own careers page before adding it to your list
Compare one verified job description with a redacted copy of your resume. Ask AI to show strong matches, partial matches, and missing requirements
Tailor the resume and write a short cover letter using only facts you can support. Review every new claim before saving or applying
Practice eight interview questions, one at a time, then add the application and next follow-up date to your tracker
Pro tip: Ask AI to show the evidence behind every resume change. A keyword is useful only when it describes work you’ve actually done.
♟️ Connect the dots. Know your next move.Visit memoket.ai
With Memoket, you can:
Connect client discussions and project updates to understand the full context
Compare conversations to spot changing needs and conflicting commitments
Ask for next steps grounded in past discussions, with sources to check
🤖 OpenAI’s agents now handle half its long research tasks
Image source: OpenAI
The Rundown: OpenAI just released new data on how AI agents are speeding up its own research, finding they now complete nearly half of tasks on their own that would take a human researcher two to four hours, up from about a quarter in March.
The details:
Agent use is exploding, with the median OpenAI researcher running ~$2.4K of AI agent work a day, and saying they’d hypothetically give up 40% of their team headcount to keep access.
Each researcher now effectively runs nearly six full days of AI agent work every day, up 11x since May, with two-thirds running four or more agents at once.
Researchers said agents still lack taste and turn out “AI slop,” but one admitted having “no confidence” that their edge in ideas will last for long.
Human researchers drove every key idea behind GPT-6 Astra, but for GPT-6.2 Astra, they said Codex helped generate key insights.
Why it matters: AI that builds better AI could make progress speed up dramatically. Two leading labs say it’s already starting, with Claude leading 26% of Anthropic’s research and OpenAI’s agents handling nearly half of mid-length tasks. Humans still call the shots, but a sudden leap could come, like the breakthroughs AI recently made in math.
▸ A reader built a personal AI health and fitness coach with Obsidian
Today’s workflow comes from an Anonymous Rundown Reader:
“I built a personal health second brain in Obsidian called Braggins, named after Bryan Johnson and David Goggins, that works as my AI fitness and longevity coach.
I store my health data as Markdown files on my laptop, including blood reports, glucose and blood pressure readings, exercise one-rep max measurements, and family health history. Every month, I export my wearable data and add it to the system. I also log meals, supplements, and sleep, and keep my strength, endurance, and longevity goals in separate files. A Focus Now doc lists what I need to prioritize each day, deferring everything else.
I use Obsidian to browse and edit these files, then open an AI app from my terminal with this context. The AI acts as my coach, providing personalized guidance based on my health data, goals, and family history. As I add advanced blood tests and scans, I plan to build a dashboard with calculated metrics such as biological age and pace of aging.”
See the workflow on our Community Workflow Hub →
📣 Marketers, we want you! What’s one thing AI takes off your plate? Content, social, reporting — anything counts. Share your workflow for a chance to be featured in front of our 2M readers.
🛠️ Trending AI Tools
⚡ Teleport Beams - Trusted runtimes for agents, access to your infra without secrets, your inference and harness*
🌎 GPT-Live-1.1 - OpenAI’s multilingual voice model, now available via API
🖼️ Ideogram 4.6 - Ideogram’s 4K multi-step image generation and editing model
📽️ Enact - Pika’s experimental AI tool that works like a live film set
*Sponsored Listing
📰 Everything else in AI today
Apple is reportedly developing a Muse-like AI agent that could control devices, use apps, browse the web, and write code, with a launch eyed for as soon as late 2027.
Google DeepMind researchers mapped four paths from AGI to superintelligence in a new essay, estimating AI compute is growing 10x a year, enough to cut what costs $1B in AI work today to about $1K within a decade.
Google selected Blue Origin to launch solar-powered AI satellites in 2029 for Project Suncatcher, its plan to build space-based data centers using proprietary AI chips.
Meta employees reportedly raised concerns that the company’s Muse AI agent was mixing up chats and exposing user data, including other users’ bank balances.
Alljoined unveiled AIR-O, an AI that uses noninvasive brain data to generate real-time images of what a person is focusing on, claiming it can adapt to new users in 15 minutes.
OpenAI is expanding its proactive dots agents to select ChatGPT Plus users and all Business Standard users, while testing a redesigned mobile app with customizable tabs.
🎓 Highlights: News, Guides & Events
Google’s shipping spree continues with Neon
OpenAI’s newest release lands to rough reviews
Zach Mink • 8 minutesAI
Rowan Cheung • 4 minutes
Stay Ahead on AI.
AI 热点