随着Anthropic(安思元)今日正式登陆公开市场,其首次公开募股(IPO)有望为公司及其投资者筹集数十亿美元。但最大的惊喜可能在于,这笔资金中究竟有多少最终会流向供应商兼投资者。
其他几家AI基础设施提供商也处于受益位置。Nvidia、AMD、Microsoft、Broadcom、Micron和TSMC已加入Google,以IPO价格向Anthropic私人配售承诺投资260亿美元。
今天的IPO标志着该融资体系的一个关键转变,即它可能变得自我维持的时刻。这些投资似乎已经得到了回报,并且现在实现了流动性。Anthropic每个季度仍处于净亏损数百亿美元的状态——但它不再需要由供应商和私人投资者买单。现在,资本将来自公开市场投资者和公司债券的组合,这两者都不存在之前引发担忧的利益冲突。
一旦上市,Anthropic预计将拥有约2200亿美元的现金、现金等价物和短期投资,其中包括其IPO收益和260亿美元的私人配售所得。Anthropic还有能力用股票支付未来的承诺。
SemiAnalysis代币经济学团队负责人Joey Brookhart估计,Anthropic在未来四年内需要大约2800亿美元的累计收入来支付其计算费用,Brookhart表示,鉴于其明年的收入前景,“这是可以实现的”。
“股权非常值得,这些公司不再需要来自[供应商]的任何支持,”Brookhart告诉TechCrunch。“Amazon不再需要扶持它们了。所有收入都来自其他客户。”
另一方面,Anthropic在2026年第三季度产生了170亿美元的收入,使其全年年化率达到680亿美元。如果Anthropic继续按照目前的轨迹增长,Brookhart相信它可以轻松覆盖资助其计算所需的缺口。
D.A. Davidson技术研究主管Gil Luria也表示同意。
“IPO完成时,Anthropic拥有的现金将比任何公司曾经拥有的都要多,”Luria告诉TechCrunch,指出该公司筹集了约1410亿美元的资金,并在每股83美元的价格下将获得约917亿美元的净收益。“而且它们并没有在烧钱。它们只是拥有一个战争金库。”
Luria区分了Anthropic的供应商投资和像CoreWeave这样的新云服务商,他认为后者更多地纯粹由循环融资支持。“如果没有Nvidia的支持,CoreWeave就不会存在,如果Nvidia不救助它们,它可能也不会存在,”Luria说。
对于这两位分析师来说,Anthropic面临的最大风险仅仅是它会输掉AI竞赛。目前,该公司的增长是由其在生成式AI前沿的地位所驱动的——但如果它被OpenAI、Google DeepMind甚至廉价开源权重模型超越,情况可能会迅速改变。正如Luria所说:“如果我们都在使用别人的模型,那么Anthropic就不值一万五千亿美元。它一文不值。”
“对于Amazon和Google来说,风险实际上并不高,因为在这种情况下,我们使用的是其他模型,”Luria说。“我主张我们将使用更多的AI。我不知道是否全是Anthropic的AI……那仍然将来自由Amazon、Google和Microsoft管理的数据中心。”
查看原文
Arriving at public markets today, Anthropic’s IPO is poised to raise billions for the company and its investors. But the biggest surprise may be just how much of that money will end up going to supplier-investors.
Amazon and Google, two of Anthropic’s largest infrastructure partners, hold stakes valued around a combined $307.9 billion at the company’s updated IPO price of $83 per share. Amazon, which invested $18 billion into the AI startup, would see its 14.7% stake hit upwards of $159 billion, while Google’s 13.7% stake is valued at nearly $149 billion, according to share ownership figures disclosed in Anthropic’s IPO prospectus.
A handful of other AI infrastructure providers are also positioned to benefit. Nvidia, AMD, Microsoft, Broadcom, Micron, and TSMC have joined Google in committing $26 billion to Anthropic at its IPO price in a private placement.
In total, Google has invested $12.3 billion into Anthropic since 2023, with another promised $30 billion dependent on Anthropic meeting certain performance milestones. And with the exception of TSMC, Anthropic has announced multibillion-dollar strategic partnerships with its other infrastructure supplier investors — including Nvidia, AMD, Broadcom, Microsoft, and Micron — that would see Anthropic purchasing their products.
This is the circular financing that has made some observers so skeptical of the AI boom. These companies are not just investors; they’re also Anthropic’s biggest suppliers. It might look like Amazon and Google are putting billions into Anthropic, but they get that money right back when Anthropic needs to train a model or serve Claude to users. From afar, it looks like unprecedented revenue growth for cloud companies and unprecedented fundraising from Anthropic — but it’s just the same money passing from one pocket to another.
Today’s IPO marks a critical change in that financing system, the moment it might just become self-sustaining. Those investments appear to have paid off, and are now liquid. Anthropic is still operating at a net loss of tens of billions of dollars each quarter — but it will no longer be suppliers and private investors footing the bill. Now, capital will come from a mix of public-market investors and corporate bonds, neither of which have the conflicts-of-interest that caused concerns before.
And critically, the company’s new bottom line looks pretty good.
Once it debuts, Anthropic is expected to have about $220 billion in cash, cash equivalents, and short-term investments, including proceeds from its IPO and the $26 billion private placement. Anthropic also has the ability to pay for future commitments in stock.
Joey Brookhart, who leads the Tokenomics team at SemiAnalysis, estimates that Anthropic needs roughly $280 billion in cumulative revenue over the next four years to pay for its compute bill, which Brookhart said is “achievable given where they will be next year from a revenue perspective.”
“The ownership stake is very worthwhile, and those companies no longer need any more support from [suppliers],” Brookhart told TechCrunch. “Amazon doesn’t need to prop them up anymore. The revenue is all coming from other customers.”
That doesn’t mean the server bills are going down, though. Anthropic has already committed to spend $900 billion, primarily for compute and infrastructure, according to a regulatory filing. Certain agreements have termination provisions, but Brookhart thinks the company is in a strong position to pay up.
On the other side, Anthropic generated $17 billion in third-quarter 2026 revenue, putting the company on pace for a $68 billion annualized rate for the year. If Anthropic continues to grow at its current trajectory, Brookhart believes it could easily cover the gap needed to fund its compute.
Gil Luria, head of technology research at D.A. Davidson, agreed.
“By the time the IPO is done, Anthropic’s going to have more cash on hand than any company has ever had,” Luria told TechCrunch, noting that the firm raised around $141 billion in capital and will take home around $91.7 billion in net proceeds at $83 per share. “And they’re not burning cash. They’re just going to have a war chest.”
A public listing will make Anthropic a more credible borrower, making it easier to raise money outside of public markets, says Brookhart. That’s despite recent reports that creditors might not see Anthropic as “investment-grade” material.
Luria distinguished between Anthropic’s supplier investments and neoclouds like CoreWeave, which he sees as more purely supported by circular financing. “CoreWeave wouldn’t exist if Nvidia didn’t prop them up, and it probably won’t exist if Nvidia doesn’t bail them out,” Luria said.
Brookhart also pushed back on describing Anthropic’s financing as purely circular, saying it was still “additive to the ecosystem.”
For both analysts, the biggest risk to Anthropic is simply that it will lose the AI race. Right now, the company’s growth is fueled by its status at the frontier of generative AI — but if it’s overtaken by OpenAI, Google DeepMind, or even cheap open-weight models, that could change fast. As Luria put it: “If we’re all using other people’s models, then Anthropic’s not worth a trillion and a half dollars. It’s worth nothing.”
But regardless of who wins the frontier, compute suppliers will likely be just fine. Neither Brookhart nor Luria believe Anthropic’s growth is at risk, despite some arguments that cheaper tokens and more competition will directly affect the AI company’s bottom line.
“For Amazon and Google, the risk isn’t actually that high because in that scenario, we’re using other models,” Luria said. “I’m arguing we’re going to use a lot more AI. I don’t know if it’s all Anthropic’s AI…. That’s still going to come from a data center managed by Amazon, Google, and Microsoft.”
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.