Henry from the Monastery of Phineas
QbitAI is public.
Even if you take a flying knife with your bare hands, you will get big results!
To be honest, at first glance, I don't think anyone would have imagined that the biggest IPO in A's history could have gone this far, and Dario Amodei's head could have gone this far.
Just as Claude's parent company Anthropic's IPO was officially opened, the rumours of a delayed exit were cleared.
According to the plan, Anthropic plans to issue 825 million shares at a price of $70.78 per share, the largest IPO ever.64,350 millionUSD, scheduled to be traded on NASDAQ on November 19, with the stock code asANT。
At that time, non-founder directors, executives, current and former employees and advisors can sell up to 10% of the prescribed benchmark holdings, or up to a maximum of about 316 million shares, from the first day of trading.
In other words, on the first day of listing, early shareholders of A had the opportunity to raise $22.1 billion (about RM150 billion).
Similarly, shareholders who want to buy A on the same day will also face the same amount of selling pressure.
In the eyes of the general public, that $20 billion is already enough for the Bay Area entertainment industry, but in the expectations of the A's, it must be said that this is a mediocre result of a heavily scaled-down version.
Why? Let's take a look at the prospects:
On November 2, Anthropic updated its IPO filings, showing a growth of 47% in the third quarter, but a slowdown of 67% from the second quarter.
On November 5, the NASDAQ Comprehensive Index fell by about 5.5% in a single day, the largest single-day decline since April 2025.
On the same day, the bonds of AI cloud service provider CoreWeave were sold, with some bond yields rising to over 17%.
……
It can be said that just as the good days of A's IPO approached, the global AI sector has collectively fallen into panic: the kill value of the kill value, the mine explosion, the debt repayment.
In return, A's potential IPO valuation also shrank from the previously discussed $2 trillion to about $1.4 trillion, a drop of about 30 percent.
In fact, according to the full dilution calculation, this figure is already below the market value of about $1.77 trillion when SpaceX went public.
Many employees also take advantage of the day, waiting for Dagonasdack to ring their bells, running straight from A to OpenAI overnight.
There are also rumours that A has postponed its listing...
However, in today's road game, all I can say is that Lone Ranger Dario still copes with the pressure.
There's a lot of money to be made out of it, a lot of money to be made out of it, a lot of money to be made out of it, a lot of money to be made out of it, a lot of money to be made out of it.
I, Dario, have to take it.
The whole world is stepping on brakes, and Dario is stepping on the gas door.
Despite the sharp decline in valuations, the growth rate of Anthropic has been enough for Chinese customers to fit into the history of technology.
Anthropic was founded in January 2021 by Dario Amodei, former Vice President of OpenAI Research, his sister Daniela Amodei, and several former OpenAI researchers.
According to PPT, since the official launch of its commercialization in the first quarter of 2023, until the third quarter of 2026, Anthropic is expected to be fully operational.A quarterly revenue is close to $17 billion.
Interestingly, the $17 billion is just a slope of the index curve if the company A gives the pie.
In addition, there is knowledge work estimated to be worth over $40 trillion a year, as well as more fields such as life sciences and robotics.
While it's a bit over the top, I have to admit that in less than six years, from an AI startup to the largest IPO ever, A has indeed contributed to growth close to the index level.
According to the issuance price cap and post-issuance basic equity, A's equity value is approximately US$1 trillion, equivalent toAbout four alibabas。
Based on the full dilution of potential shares including options and restrictive stocks, the valuation is about $1.4 trillion, second only to the largest IPO ever since the merger of xAI and SpaceX.
A company's IPO totals 825 million shares, including about 801 million newly issued shares, and about 23.65 million shares sold by existing shareholders.
Here, it's really worth mentioning, a bunch of math-savvy dads ready to step out of the closet.US$26 billion in private funding.
These include $10 billion in Google, $2.5 billion in AMD, $1 billion in Bloomberg, $1 billion in AML, and $500 million in private equity investments.
In other words, the IPO itself raised a maximum of $64.35 billion, plus the $26 billion in strategic private funding, bringing the total funding to close to $90.4 billion.
Of course, as to how the money would end up, and at what price, it would go into the pocket of Mr. Gold's father, to see what Dario would do next.
In terms of operational data, as we said at the beginning, A.Income growth remains exaggerated, but computational spending is still expanding further, and the growth curve has been slowed as a result.
In the third quarter, Anthropic posted revenue of $169.56 billion, 12.27 times the year-over-year average, growing at a compound annual growth rate of about 47%, but slowing significantly compared to the second quarter.
Looking at the full year, A's cumulative net income of $33.24 billion for the first nine months of the year, the annual operating income disclosed on October 31 was about $81 billion.
At the same time, customers who contributed at least $100,000 in revenue in the past 12 months have surpassed 119,000. Compared to more than 950 at the end of September 2025, this is an increase of more than 12 times a year.
Of these, more than 1,900 customers contributed more than $1 million in annual revenue and about 250 customers contributed more than $10 million.
What's even more exaggerated is that the retention rate of net income from stock customers has reached about 900% according to the A company caliber.
This also means that Anthropic's growth is not entirely dependent on continuing to innovate, but on continuing to exploit the paying potential of existing enterprise customers.
From this point of view, Dario did come up with a pretty good commercial score.
But the problem is right here.
As the two days of panic on Wall Street reflected, the market is now less suspicious of the use of AI, but has begun to take another bill seriously:
Is the revenue generated by AI enough to cover the huge amount of computing expenditure and data center investment behind it?
For Anthropic, expensive purchases of computing power are increasingly costing this near-index growth curve.
In the third quarter, Anthropic's computing and infrastructure expenses reached $98.63 billion, up from 48% to 58% of revenue in the previous quarter.
The company explained that of the newly added deliveries for the quarter, which included about $5 billion in short-term resources, the price was significantly higher than the similar resources purchased through multi-year agreements.
The usual explanation is that because demand is growing too fast, A companies have to pay a premium for short-term computing power that can immediately become a force.
As a result, A's adjusted operating profit increased from $22.35 billion to $2.46 billion in the previous quarter, while its adjusted operating profit dropped from 19% to 15%.
According to the U.S. General Accounting Standards, seasonal operatingThe loss was $567 million.
In addition to the current cost, A and Dario have a huge bill for the next decade.
It is reported thatIn order to support the next growth, A has already signed a related procurement and lease pledge of about $900 billion in advance.
Of these, about $45.16 billion is expected to occur before the end of 2030, with the remaining majority between 2031 and 2045; about $70.7 billion will need to be paid in the 12 months following the end of September 2026.
In other words, while a business that can continue to produce blood in addition to coding needs further verification, A.S.A. has already arranged the machines and chip resources for the next few decades.
Whether it's early spending or continuing to scale the index, at least today, is still a question number.
How can we reduce our productivity?
In this road trip, Anthropic also demonstrated its own reduction-increased efficiency plan.
On the one hand, using a combination of multiple chips such as AWS Trainium, Google TPU, Invidia and AMD GPU, matching more appropriate algorithms according to the task;
On the other hand, by leasing chips directly through Anthropic Direct, deploying data centers reduces long-term computing costs.
At the same time, it has been revealed that the amount of code that Anthropic engineers contribute daily has reached 45 times the average level in 2024.
Prior to the AMD GPU, the company even had Mythos translate and write almost all of its underlying computing cores in three days.
It's fair to say that Dario has taken the lead in getting AI involved in the company's own R&D and production processes.
AI will reduce training and R&D costs while allowing AI to generate more revenue.
Company A is also moving forwardCopying the proven business models of coding into more knowledge-based work scenarios.
For example, just today, Claude released a series of office products, including Claude Sheets, Claude Whiteboards and Claude Data.
In addition to connecting to third-party applications such as Adobe, Notion, Claude is trying to connect work like formwork, data analysis, collaboration, and content production.
In the short term, office AI will be the centerpiece of A's bet, while in the long run, there will be larger markets like life sciences, robotics and more.
However, as the story goes, Wall Street still seems to have its own idea.
CNBC's comment section on the ongoing IPO is blank:
Dario is also working on the future of CardAI, and the IPO has been blocked.
Next week, he will know if he is a murderer, or if he keeps the clouds open and sees the moon.
By the way, according to Bloomberg, Dario is also scheduled to have a live chat with GM CEO Jamie Dimon tomorrow.
Reference links
[1]
