Zhitong Finance App learned that, according to reports, Anthropic plans to host a pre-IPO Investor Day event at its San Francisco headquarters on October 14 and disclose its third-quarter revenue data to a group of potential institutional investors. In a context where Anthropic is advancing its listing plan, this latest revenue data is expected to allow the market to understand its growth and profit situation and assess whether the $2 trillion valuation it is seeking is reasonable. At the same time, it may also have a huge impact on artificial intelligence (AI) transactions centered on the technology sector.
It is worth mentioning that Anthropic's disclosure comes at a time when investors are increasingly scrutinizing the financial performance of private AI companies to assess the sustainability of the entire industry's growth. Last week, a report disclosing OpenAI's revenue and accounting treatment methods caused confusion and dragged the AI sector of the US stock market collectively downward. OpenAI revealed to investors that by the end of September, the company's annualized revenue scale was about 50 billion US dollars, lower than the 70 billion US dollars widely reported in the market at the end of last month. Although people familiar with the matter said that this updated figure excludes revenue from cloud partners and does not reflect any changes at the demand level. It's unclear whether Anthropic will use the same statistical caliber, that is, whether partner revenue will be included.
The IPO has entered a critical stage! Explosive revenue growth supports sky-high valuations
Anthropic's listing plan is gradually becoming the focus of the market. According to recent reports, Anthropic may officially launch an IPO promotion as early as November 9, so it is expected that trading will begin before Thanksgiving on November 26. According to reports, the company is seeking a valuation of 2 trillion US dollars. If this valuation is achieved, Anthropic will surpass SpaceX's issuance valuation of 1.77 trillion US dollars when it went public in June this year, and set the record for the largest IPO in global history. It is important to note that $2 trillion should currently be viewed as a potential valuation in market discussions, rather than an established offering price or formal company commitment.
Anthropic, an AI upstart that has only been in business for five years, is rushing from a private unicorn to the open market at an astonishing speed on Wall Street, and the most direct driving force comes from the amazing explosive power of its financial data.
By the end of July 2026, Anthropic's annualized revenue operating rate (ARR) had exceeded US$65 billion. This figure is more than 7 times higher than the level of about 9 billion US dollars at the end of 2025, and an increase of 38% from the 47 billion US dollars in May. In the second quarter of 2026, the company's initial revenue exceeded US$11.5 billion, compared to only US$787 million in the same period in 2025 — an increase of more than 14 times over the previous year. What made the market even more excited was that Anthropic achieved a positive adjusted operating profit of approximately US$559 million in the second quarter.
This growth trajectory is steep: ARR was only about US$87 million at the beginning of 2024, US$1 billion at the end of 2024, US$9 billion by the end of 2025, surpassing US$30 billion in April 2026, reaching US$47 billion in May, and rushing to US$65 billion at the end of July. Investors' most aggressive predictions are that annualized revenue will reach $100 billion to $120 billion by the end of 2026.
Behind the explosive increase in revenue is Anthropic's clear commercialization path. Unlike OpenAI, which relies on C-side subscriptions, over 80% of Anthropic's revenue comes from API call charges from enterprises and developers. The number of corporate customers has exceeded 300,000, and about 70% of the Fortune 100 companies have achieved deep adoption. The number of customers spending more than $1 million a year has doubled in recent months to more than 1,000. The AI programming tool Claude Code is the core driving force. ARR surpassed 2.5 billion US dollars in early 2026, and it is estimated that ARR had reached 15.12 billion US dollars by August.
Significant improvements in reasoning efficiency have further strengthened the positive cycle of the business model. Using an API-centered, pay-per-usage model, Anthropic's comprehensive gross margin rose to the middle of 60%, and inferred gross margin rose to 70% to 85% from 38% a year ago. For cutting-edge model companies that are still in the stage of large-scale computing power investment, this is a key sign that the scale effect is beginning to show operating leverage.
Bank of America Securities's chief strategist pointed out that the $2 trillion valuation implied revenue expectations for 2028 and required a growth rate close to three digits. According to reports, Anthropic internally expects revenue of about 190 billion to 200 billion US dollars in 2028. If we calculate the revenue of 100 billion US dollars at the end of 2026, the forward-looking market sales ratio corresponding to the 2 trillion US dollar valuation is about 20 times, which is lower than the market sales rate of popular AI stocks such as Palantir and Nebius, which can easily be more than 50 times that of popular AI stocks such as Palantir and Nebius. Some investors even assert, “If Anthropic can maintain an annual growth rate of 800%, the company's valuation would be close to $3 trillion, even with a conservative 30-fold revenue multiple.”
Notably, the other side of high growth is huge cost investment. Anthropic's net loss in 2025 was close to 42 billion US dollars, which is more than 5 times the previous year's net loss of about 8.3 billion US dollars. However, about $34 billion of this was due to changes in the fair value of convertible financing instruments — accounting losses due to soaring valuations, not cash expenses. Excluded, the actual operating loss was approximately US$8.06 billion, an increase of 170% over the US$2.98 billion in 2024.
The reason for the increase in losses was the expansion of upfront expenses for computing power. In 2025, spending on computing power and infrastructure reached US$7.33 billion, equivalent to 1.6 times that year's revenue. But the loss rate is narrowing — the operating loss rate fell from around 750% in 2024 to 176% in 2025.
The $2 trillion valuation reflects investors' high expectations for the AI industry's long-term growth space. Compared with traditional software companies, Anthropic not only hopes to earn revenue through model subscriptions, enterprise services, and API interfaces, but also bets that AI can gradually enter core enterprise scenarios such as programming, customer service, knowledge management, and business process automation. As enterprises move from experimental applications to production deployment, there is an opportunity to upgrade AI services from ancillary tools to part of the enterprise infrastructure.
In particular, the expansion of the Claude series model in the field of enterprise applications and programming has enabled Anthropic to seek more sustainable commercial demand. Once enterprise customers embed AI models into software development, internal knowledge retrieval, or daily operation processes, they often need to continue to pay for model calls and services. This kind of demand is expected to generate high revenue stickiness, and also provide support for the company's future revenue growth.
However, high valuations also mean higher fulfillment requirements. Investors are not only buying Anthropic's current size of business, but also paying for its growth expectations over the next few years. If revenue growth slows down after listing, profit improvements fall short of expectations, or if the AI industry's overall valuation shrinks, the company's stock price may face significant pressure. Therefore, whether a $2 trillion valuation can be established ultimately depends on whether the market believes in Anthropic's ability to transform current technological leadership and demand growth into long-term cash flow generation capabilities.
$518 billion infrastructure commitment! Growth engines are also sources of financial stress
If revenue growth is at the core of Anthropic's valuation story, then the huge demand for computing power and infrastructure is one of the cost factors that need to be examined most in this business model.
According to reports, according to Anthropic's IPO prospectus seen by the media, the company plans to assume approximately $518 billion in cloud computing and infrastructure-related expenses or commitments over the next few years. This scale shows that the company's judgment on future AI needs is very aggressive, which also means that its commercial expansion will be deeply tied to cloud computing services, data center construction, and advanced chip supply.
It is important to note that this figure is a commitment to related expenses and infrastructure for many years to come. It does not mean that the company needs to pay the full amount in one time in a certain fiscal year, nor can it simply be equated with capital expenses or cash outflows that have already occurred. But even when considering differences in execution times and contract structures, such long-term commitments can still have an important impact on a company's future financial flexibility.
There is a clear difference between the cost structure of AI companies and traditional software companies. After completing product development for traditional software, the marginal delivery cost for adding a new user is usually relatively limited; however, large model services require continuous inference calculations. The more frequent user calls and the more complicated the model processing tasks, the higher the related computing power costs. For products that can perform complex tasks over a long period of time, such as programming agents, increased user usage may not only bring more revenue, but also drive up the consumption of computing resources.
Therefore, whether Anthropic can establish attractive profit margins in the future depends on whether multiple factors can be improved at the same time: whether model reasoning efficiency can be improved, whether unit calculation costs can be reduced, whether corporate customers are willing to accept higher prices, and whether the company can reduce computing power procurement costs through long-term contracts and resource scheduling.
If the commercial value of AI models continues to increase and enterprise customers are willing to pay for higher productivity, then huge infrastructure investment may become a necessary condition for establishing competitive barriers. Conversely, if model capabilities gradually converge, service prices continue to decline, or if enterprise customers begin to reduce AI budgets, long-term computing power commitments may become a heavy cost burden.
This also means that the market cannot simply view the $518 billion infrastructure commitment as a sign of benefit to the chip and data center industry. For upstream suppliers, it may represent future demand; for Anthropic itself, it also represents performance obligations, resource dependency, and future profit pressure.
With listing valuations already heavily dependent on future growth, Anthropic needs to prove to investors that these expenses are not simply investments chasing model size, but long-term investments that can bring sufficient commercial returns.
Customer concentration and cloud platform dependency or limit revenue quality
Another important risk for Anthropic comes from the revenue structure. According to reports, Anthropic's prospectus shows that in 2025, the company received revenue of 2.16 billion US dollars through Amazon and Google's cloud platforms, accounting for 47% of the annual revenue. This ratio was only 11% in 2023 and rose to 32% in 2024, indicating that the company's dependence on cloud platform channels is gradually deepening.
For Anthropic, the cloud-based sales model can help the company significantly reduce the difficulty of acquiring customers. Many large customers purchase Claude services directly through the cloud marketplace, which can reduce contracts, settlement, and technology deployment. Anthropic stated in its prospectus that these cloud platforms have sales networks that are difficult for a single enterprise to replicate independently, which helps Claude enter the global enterprise market faster.
The problem, though, is that Amazon and Google aren't ordinary channel providers: they both invest in Anthropic, sell the computing resources needed for training and inference, and have their own AI models and cloud computing businesses. This means that these competitors can access some of Anthropic's pricing and commercial terms, and may affect computing power allocation, product promotion, and customer relationships. Anthropic also acknowledged in its prospectus that reliance on a few partners and suppliers could cause conflicts of interest and adversely affect the company's access to computing resources.
A high customer concentration doesn't necessarily mean there's a problem with the business model. For AI companies in the rapid expansion stage, large technology companies, enterprise customers, and cloud platform partners can often provide a considerable procurement scale in the early stages, which also helps push products into a wider range of application scenarios.
However, when a small number of customers contribute a large percentage of revenue, the company's sensitivity to customer procurement decisions also increases. Anthropic's revenue growth and profit margins may be affected once key customers adjust procurement budgets, switch to self-developed models, reduce call volumes, or demand lower prices in contract negotiations.
What's more complicated is that some of the big partners in the AI industry are both customers and computing power suppliers or investors. While this relationship helps Anthropic gain access to computational resources and commercial channels, it may also cause the company to face some degree of dependency. If partners change commercial terms, adjust resource allocations, or use more resources to promote their own models, Anthropic will need to address potential cost and competitive pressures.
The cloud platform channel also involves the difference between revenue recognition and financial benefits. Sales revenue obtained through partners can expand an enterprise's customer reach, but a portion of it needs to be paid to cloud service providers. As a result, investors need to pay attention to revenue, gross profit margin, customer renewal rate, and customer retention after deducting channel sharing, not just the unadjusted operating rate of revenue.
After listing, Anthropic's ability to continue to expand its direct corporate customer base, reduce its dependence on a small number of customers, and improve the diversity of revenue sources will be an important indicator for measuring the resilience of its business model.
AI competition is accelerating, and technological leadership may not necessarily translate into excessive profits in the long term
The competitive environment Anthropic is facing cannot be ignored either. Competition in the generative AI market comes not only from direct rivals such as OpenAI, but also from self-developed models, open source models, and emerging vertical solutions from major technology companies. As model capabilities continue to improve, enterprise customers are increasingly likely to comprehensively consider performance, price, data security, system compatibility, and deployment costs when choosing AI services, rather than simply relying on the brand advantages of a model.
For Anthropic, the advantages of the Claude series model in enterprise services and programming scenarios are an important foundation for it to acquire customers. However, the model industry is characterized by rapid iteration, and today's technological leadership may not continue for the next few years. Even as companies continue to launch new models, they must continue to invest in R&D, computing power, and talent to maintain competitiveness. According to reports, the prediction market is currently betting that Google (GOOGL.US) may overtake Anthropic and lead the 2026 AI model competition.
Furthermore, improvements in model capabilities do not always go hand in hand with commercial profits. Stronger models may help companies solve more complex problems and increase customer willingness to pay; but if competitors provide close performance at lower prices, Anthropic may face price reduction pressure. If inference costs fall faster than sales prices fall, profit margins may improve; conversely, price competition may offset the benefits of increased efficiency.
The development of AI agents has provided Anthropic with new opportunities for growth. Compared to chatbots that simply answer questions, agents that can perform programming, research, and business process tasks may generate more clear enterprise productivity benefits, and also have the opportunity to drive higher frequency of use and deeper workflow integration.
But agents are also likely to introduce higher computational consumption, reliability requirements, and security responsibilities. The fact that companies are willing to pay for AI that can complete tasks does not mean they are willing to bear unlimited calling costs. Anthropic needs to prove that the commercial value of smart products can cover their resource consumption and form a replicable profit model.
Therefore, technological leadership is only part of Anthropic's valuation. Longer term competitive advantage also comes from various aspects such as customer relationships, enterprise workflow integration, product reliability, model efficiency, and cost control.
Legal conflict with the US government adds difficult to quantify policy risks to IPOs
In addition to financial and competitive issues, Anthropic also faces a particular risk closely linked to corporate governance, business development, and technology security: disputes with the US Department of Defense over restrictions on the use of AI.
According to reports, Anthropic has requested the US Federal Court of Appeals in Washington, D.C., to reconsider the previous ruling upholding the US Department of Defense decision. Previously, the court supported the listing of Anthropic as a national security supply chain risk company with a 2-1 result. The core of the controversy is that Anthropic refused to allow its AI technology to be used for autonomous weapons or large-scale surveillance of the American public, and the US Department of Defense believes that related restrictions may affect military operations.
Anthropic, on the other hand, argued that the relevant government decision went beyond the scope of legal authorization and was in retaliation against its AI security stance. The company said earlier that this determination has led to billions of dollars in business losses. Meanwhile, another related legal dispute unfolded differently in August of this year: a federal judge in California overturned a parallel ruling based on another law. It can be seen from this that the relevant legal issues are still complicated and cannot simply be regarded as having settled.
The impact of this conflict on Anthropic is not limited to potential government contract revenues. Government decisions may affect their eligibility to participate in federal projects, cooperation with government contractors, and some clients' judgments about supplier compliance risks. Even if the company has strong demand in the commercial enterprise market, policy risks may change the purchasing decisions of some customers.
At the same time, Anthropic's position also relates to its brand positioning and long-term strategy. The company has always emphasized AI safety and is more cautious about the risks and regulatory issues of advanced AI systems. If it relaxes relevant restrictions to seek government business, it may raise questions about corporate principles and security commitments; however, if it insists on strict restrictions, it may also lose some government contracts or face adverse effects at the policy level as a result.
From a capital market perspective, this is a type of risk that is difficult to measure directly through traditional financial indicators. It could impact revenue as well as the company's reputation, regulatory environment, and future business boundaries. Listing documents need to fully disclose relevant lawsuits, potential losses, and their impact on business, and investors also need to assess whether these disputes are short-term uncertainties or structural factors that may affect the company's growth path in the long term.
Anthropic's listing will be an important stress test for AI deals
Anthropic's listing process is not only related to the financing and shareholder withdrawal of an AI company, but may also become an important window for the market to test the entire AI industry's business model.
If the company can continue to prove rapid revenue growth after disclosing financial data, while improving profit margins, controlling infrastructure costs, expanding the customer base, and reducing policy risks, then its IPO may further strengthen the market's confidence in the prospects of commercializing enterprise-level AI and bring new demand expectations to the AI infrastructure and chip industry.
However, if financial disclosure shows that revenue growth is highly dependent on a small number of customers, that cloud platform channel costs are high, or that long-term computing power promises are putting significant pressure on cash flow, then investors may re-examine their valuation basis. If legal disputes with the government further affect the business, it may also cause the market to demand higher risk discounts on its long-term growth path.
For potential investors, what Anthropic is most concerned about is not a single valuation figure, but the relationship between revenue, cost, profit, cash flow, and risk. Rapid growth can prove market demand, but it cannot automatically prove that the business model is mature; huge infrastructure investment can support future competition, but it may also amplify financial pressure when demand falls short of expectations.
At the end of the day, Anthropic is facing one of the core questions of the AI industry: as big models move from technological breakthroughs to large-scale commercial applications, can enterprises establish sustainable economic returns while continuing to innovate.
The $2 trillion valuation represents not only investors' expectations for Anthropic, but also a promise that must be fulfilled by future results. What really determines the success of its IPO will not be how enthusiastic the market is about the AI vision, but rather whether the company can prove that revenue growth can ultimately be transformed into stable profits, sustainable cash flow, and competitive barriers sufficient to support long-term valuations.