ipo

Anthropic, the artificial intelligence company behind Claude, could be preparing for one of the biggest technology IPOs ever.
Recent reports say Anthropic expects its eventual public offering to match or even exceed the size of SpaceX’s record breaking IPO. Bloomberg reported that the company is running the numbers as it prepares for a possible public filing.
The important point is that this is not confirmed yet. Anthropic has confidentially submitted a draft registration statement to the US Securities and Exchange Commission, but it has not announced the final IPO size, share price or valuation.
SpaceX Set a Huge Benchmark
SpaceX made history in June 2026 when it priced its IPO at $135 per share.
The offering initially raised about $75 billion and valued SpaceX at approximately $1.77 trillion. After the overallotment option, the total amount raised increased to about $86.2 billion.
That gives Anthropic a very high target to beat.
Why Anthropic Has a Chance
Anthropic has experienced extraordinary revenue growth.
The company was valued at $965 billion during its $65 billion funding round in May 2026. Its reported annualized revenue run rate reached about $65 billion by July, compared with roughly $9 billion at the end of 2025.
Some investors reportedly expect Anthropic’s annualized revenue to reach $100 billion to $120 billion by the end of 2026.
That growth is one of the main reasons investors are discussing a valuation around $2 trillion or potentially higher.
But There Is a Big Difference
A huge valuation does not automatically mean a huge IPO.
For Anthropic to beat SpaceX’s roughly $86 billion final IPO proceeds, it would need to sell enough shares at a price that investors are willing to accept.
At the same time, the company must convince public market investors that its rapid growth can continue.
This is where the story becomes more complicated.
Anthropic is spending enormous amounts on computing infrastructure, model development and talent. Reuters reported that its valuation depends heavily on expectations for much larger future revenue, including a possible $190 billion to $200 billion revenue level in 2028.
The Main Risks
The AI market is extremely competitive.
Anthropic faces powerful rivals including OpenAI, Google and several open source AI projects. Cheaper AI models could also put pressure on pricing and margins.
There are also questions about infrastructure costs, regulation and the long term profitability of frontier AI companies. The Financial Times has warned investors to look beyond revenue growth and examine margins, competition and the rising cost of developing advanced models.
Another concern is valuation.
A $2 trillion valuation would put Anthropic in an extraordinary position for a company that is still building its long term profit model. Investors would need to believe that its future earnings can justify such a price.
My Take
Anthropic could break SpaceX’s IPO record, but it is too early to say that it will.
The strongest evidence is its explosive revenue growth, strong demand for Claude and massive investor interest in AI.
The biggest question is whether that growth can translate into sustainable profits.
If Anthropic reaches the reported $100 billion to $120 billion annualized revenue range and public markets remain enthusiastic about AI, a record setting IPO is possible.
But investors should remember one simple rule: a record IPO does not automatically mean a good investment.
The final IPO valuation, share price, amount raised and financial disclosures will matter much more than the headlines.
For now, Anthropic is clearly positioning itself for a historic public debut, and the race between AI and space for the title of the biggest IPO has become one of the most interesting stories in the 2026 market.
$ANTHROPIC #AnthropicIPOCouldTopSpaceXRecordReportsSay #IPO #SpaceX

Unitree Technology has announced its initial public offering (IPO) price at 150.80 yuan per share, according to the Shanghai Stock Exchange. The company revealed that, following the offering, its total share capital will stand at 404 million shares.
Of these, approximately 30.0877 million shares will be unrestricted and tradable when the company begins listing, representing about 7.44% of the total capital. The IPO price corresponds to a 2025 diluted static price-to-sales ratio of 35, indicating how the market values the company's sales relative to its current stock price.
This IPO marks a significant milestone for Unitree Technology as it seeks to raise capital and expand its operations. The company's valuation and share structure are now set to be evaluated by the market once trading commences, providing investors with an opportunity to participate in its growth prospects.
The offering comes amid a broader environment of increased activity in Chinese tech and robotics firms seeking to list and raise funds. The company's success in attracting investor interest will likely depend on its growth potential and the outlook for its industry segment. #IPO #Unitree #StockMarket

Anthropic Q2 营收115亿美元，同比14倍。
数字非常炸裂。
但炸裂的数字最该问一句：为什么偏偏现在不报ARR了？
记住一点统计口径一旦变了，那一定是在掩饰什么？
那他们到底在掩饰什么？
年初到五月底，ARR公布得那叫一个欢。
4月300亿，5月470亿，恨不得天天发捷报。
然后6月1日秘密递交S-1，进入静默期，ARR再没公布过。
静默期是法律要求，没毛病。
但问题是：如果6月7月还在飞冲，以Anthropic之前恨不得拿大喇叭喊ARR的德性，你觉得他们忍得住不说？
第三方估7月ARR约741亿。
听着很猛。但从5月的470亿到7月的741亿，两个月涨57%。
对比一下——年初到5月是从几十亿冲到470亿，月环比翻倍。
增速在放缓。
这就是他们要掩饰的东西，
不是在跌，是在踩刹车。
这115亿到底是什么钱？
不是你在网页上聊Claude的钱。
是Claude Code的钱。
去年9月上线，年化收入从5亿起步，到今年2月25亿，5月有信源说到了80亿。
一个不到一年的产品做到这个体量，翻SaaS历史找不出几个。
Copilot帮你打字。
Claude Code帮你干活。
读你的整个项目，改文件，找bug，跑测试，执行终端命令。一个卖补全，一个卖劳动力。
编程AI市场Claude Code占54%，Codex只有21%。GitHub公开提交里4%是Claude Code写的，一个月翻一倍。
企业内部数据更夸张——合并进代码库的代码超过80%是Claude写的。
卖的不是token，是工程师的时间。
年薪20万美元的工程师，每年花几千到一万美元买AI工具，效率提升20%就回本。
企业算过账的，省下来的比花出去的多，停不下来。
这就是为什么Anthropic 80%收入来自企业API，不是消费者订阅。财富10强里8家是客户。
年消费超百万美元的企业三个月翻倍，从500家涨到1000多家。
订阅可以取消。
算过账的生产工具不会。
但钱赚到了，账不一定算得平。
调整后营业利润转正了，首次 约5.59亿美元 听着不错？
股权激励、一次性费用、折旧口径——全塞进调整项里。GAAP口径下大概率还在亏。
Anthropic自己都警告过：后续季度利润率会因为算力投入加大而承压。
下一代Mythos级别模型烧钱只会更多。
推理毛利率从38%飙到70-85%？
折旧和电费在不在毛利率计算公式里？
如果算力硬件按6年折旧但经济寿命只有2.5年，3.5年的折旧错配就是报表上的欺诈。
这跟WeWork的剧本一模一样。
上市前风光无限，S-1揭穿高增长背后的巨额亏损和不可持续的单元经济。
2万亿估值的赌注。
6月秘交S-1，摩根士丹利高盛摩根大通联合承销，10月上市。
估值从5月的9650亿被推到2万亿。
年底ARR如果到1000亿，市销率20倍。
美股高增长AI公司普遍30倍。这么看不算贵。
前提是增速能维持。
但竞争呢？
6月GLM 5.2。
7月GPT 5.6、Kimi K3。
7月底DeepSeek V4 Flash。
8月Qwen 3.8 Max。
昨天Qwen 3.8 27B——家用电脑能跑的模型，Agent Coding能力对标年初的Opus 4.6旗舰。
27B。家用电脑。Opus 4.6级别。
这玩意儿对Anthropic的杀伤不是"开源追上来了"那么温和。
是"你卖500刀的旗舰能力，别人免费送还能本地部署"。
企业不傻。会用脚投票。
微软腾讯这类大企业已经开始大砍token支出，找更廉价模型替代。
多个会计事务所发现使用AI后增长没预期那么大。
token消耗只是工作量证明，但工作量本身已经廉价化。
工作不是比特币挖矿。
不是烧电就有价值，得带脑子。
Anthropic证明了一件事：AI能赚钱。
在编程这个场景，公式成立了：客户花100块买AI，AI帮客户省300块人工，模型公司提供服务的成本60块，40块是利润。
过去整个AI行业最缺的就是这个公式的证据。
Anthropic在编程场景里把它证明了。
但这也只是编程一个场景。
客服、法务、财务、销售——能不能复制？
不知道。如果只有编程能赚钱，天花板是看得见的。
更重要的是：这个公式成立的前提是"模型能力领先"。
一旦开源追平——而GLM 5.2和Qwen 3.8正在做这件事——公式的分子会被压下来，分母不一定能同步降。
结果就是利润率被两头挤。
所以Anthropic的财务魔术能玩多久？
取决于三件事：
下一代模型能不能重新拉开和开源的差距。
能，高毛利继续。不能，变commodity。
编程之外的场景能不能跑通同样的公式。
能，天花板打开。
不能，就是niche生意。
IPO之前能不能把故事讲到让投资者相信"这不只是Opus 4.6一个季度的红利"。
目前看，Q2大概率是Anthropic最高光的一个季度。空前，也很可能绝后。
下半年竞争压力只会更大，企业token支出只会更精打细算。
以后每天都是Anthropic在未来最好的一天。
这不算看空。
算算账而已。
#Anthropic #Aİ #IPO $ANTHROPIC