Following the explosive tech headlines from Claude developer Anthropic, which revealed that its overall revenue for 2025 skyrocketed to roughly 12 times the previous year's figure and that its valuation surged past $2 trillion, ChatGPT's developer OpenAI is not to be outdone. On Tuesday, Eastern Time, the company disclosed that its annualized revenue has approached $70 billion, representing growth of more than 70% since July. Moreover, OpenAI, a global leader in large AI models and AI applications, is seeking to raise at least $30 billion at a pre-money valuation of approximately $1.4 trillion, a substantial increase from the roughly $850 billion valuation seen in its previous funding round. OpenAI is attempting to use this massive new private financing round to buy time for both its rapidly expanding AI business and stricter safety governance.
OpenAI CEO Sam Altman stated that the company wants to first adapt to the safety requirements brought about by leaps in frontier model capabilities before enduring the capital market pressures of being a newly public company. As a result, a potential IPO would not come until 2027 at the earliest. On the other side, Anthropic is targeting an IPO valuation exceeding $2 trillion, with previously discussed fundraising targets reaching as high as $100 billion. If the offering is ultimately completed at that scale, its fundraising amount would surpass SpaceX's approximately $85.7 billion including the over-allotment option, potentially setting a new global IPO fundraising record. However, Anthropic's financing plans and target valuation remain proposed arrangements and cannot be equated with the company having already set a historic IPO record. The latest news also indicates that its listing could be delayed until after the U.S. midterm elections in November.
The capital choices of these two AI large-model super-giants jointly reveal that, driven by the wave of AI agents focused on agentic automated workflows, the large-scale commercialization of global AI applications is pushing fundraising capacity, safety governance, and AI compute infrastructure supply capability into equally important positions. With the most cutting-edge AI agents represented by Muse and Astra rapidly gaining global popularity, AI is accelerating its shift from a conversational tool to automated programming, research, and the execution of complex enterprise workflows, simultaneously expanding both application revenue and the scale of AI compute resource investment. According to information from people familiar with the matter, Anthropic's IPO prospectus details also show that its operating loss for the year exceeded $8 billion, with compute and infrastructure spending reaching $7.33 billion, highlighting the scale of global AI computing resource expenditure, which is clearly showing a surge.
In the view of Anthropic's management, the memory chip components in AI data center server clusters, along with AI GPUs, remain the clearest supply bottlenecks in the AI compute industry chain. Market research firm TrendForce estimates that server DRAM contract prices will cumulatively rise by about 270% in 2026, while enterprise-grade SSD prices will cumulatively rise by about 235%. In 2027, HBM contract prices may still rise by 70% to 140%, continuing to show doubling-style growth. These figures reflect the combined effect of continuously expanding AI compute demand and rising memory chip prices. TrendForce calculations also show that shipments of NVL72 racks covering the Blackwell and Vera Rubin platforms are expected to grow by more than 50% year over year in 2027. A chart from its market research indicates that the output value of related systems is projected to rise from about $226 billion in 2026 to $711 billion in 2027, a sharp year-over-year increase of 214%.
Anthropic and OpenAI are preparing to bring this expansion into the global public capital markets. People familiar with the matter said that, in addition to showing roughly 12-fold revenue surge, Anthropic's IPO prospectus also reveals that the company will bear future obligations totaling as much as $518 billion over the coming years for cloud services, AI computing capacity, and broad AI infrastructure spending. A potential listing valuation exceeding $2 trillion means investors will directly price the commercialization capability of frontier models, compute procurement commitments, and long-term profitability prospects, with effects extending to the valuation expansion path of OpenAI and related AI compute infrastructure suppliers.
Listing can wait, but intense competition in AI product lines continues
OpenAI CEO Sam Altman said the company wants to get through this period of heightened AI safety concerns without bearing the pressures of being a newly public company, and he believes investors will remain patient with its IPO plans. In an interview with media on Tuesday, Altman said the developer of ChatGPT and Codex is currently adapting to the new level of AI capability and the additional safety requirements that come with it. "We just want to find our footing first, make sure we understand how to operate under this new paradigm, and be able to make decisions about some of the issues we need to face without the pressure of being a newly public company," Altman said. These remarks are among his most detailed explanations to date of the company's decision to delay a potential listing until as early as next year.
OpenAI and its long-time rival Anthropic have both previously submitted listing application documents confidentially, and Anthropic is expected to go public as early as this autumn. In recent weeks, after a series of cybersecurity incidents involving the two companies' AI systems and warnings from grassroots tech industry employees about existential risks to humanity, both AI companies have come under a new round of scrutiny over the potentially catastrophic harm their technology could cause. OpenAI said this week that it had canceled plans to release the GPT-6.1 version of its flagship Astra model, noting that risks have risen after a recent string of AI boundary-crossing intrusion incidents. The company also apologized to Australia this week after some of its AI models intruded into Australian government websites.
While reconsidering the timing of its IPO, OpenAI is in discussions for a new funding round. Bloomberg disclosed on Tuesday that the company plans to raise at least $30 billion from investors at a valuation of about $1.4 trillion. This round is intended to serve as bridge financing, providing the company with ample funds while it temporarily refrains from an IPO. "We will go public someday," Altman said. In addition, according to the latest information from internal sources, OpenAI is seeking to raise $30 billion at a $1.4 trillion valuation. According to media reports, people with internal knowledge said that after delaying its initial public offering plans, AI startup OpenAI is seeking to raise at least $30 billion from investors through a new funding round. These people said the company is seeking a valuation of about $1.4 trillion, excluding the proposed funds from this round, which could make its valuation once again exceed the most recent private-market valuation of long-time rival Anthropic. Bloomberg previously reported that OpenAI was considering a new funding round at a $1.2 trillion valuation. The people said fundraising discussions are still at an early stage and arrangements may change. Because the information has not yet been made public, they requested anonymity. One person said that, against the backdrop of strong revenue growth, demand for this round is mainly investor-driven. Representatives for OpenAI declined to comment on any fundraising or IPO matters.
Ahead of its planned Wall Street listing, this ChatGPT developer has been engaged in fierce competition with Anthropic to win more enterprise customers and expand revenue. Both companies have confidentially submitted listing application documents, and Anthropic is expected to go public as early as this autumn. OpenAI CEO Sam Altman recently said the company will not go public this year. In an interview with U.S. media on Tuesday, he said the company wants to get through this period of heightened AI safety concerns without bearing the pressures of being a newly public company. Altman said this ChatGPT developer is currently adapting to the new level of AI capability and the additional safety requirements that come with it. He added that he believes investors will remain patient with the company's IPO plans.
People familiar with the matter said the latest funding round is intended to serve as bridge financing, replacing a public-market IPO in providing OpenAI with additional capital. OpenAI's most recent funding round was in March, when it raised $122 billion, with a valuation of $852 billion including the funds from that round. Like its peers, OpenAI is under increasingly strict scrutiny because AI could cause catastrophic harm, including through cybersecurity intrusions. OpenAI's technology has been involved in multiple safety incidents, including the incident targeting Australian government websites. The company also said this week that it will not release its latest model, GPT-6.1 Astra, because the system did not meet internal safety standards. Altman supports a plan proposed by Anthropic CEO Dario Amodei to slow the development pace of the most frontier AI models and introduce independent evaluators to help safeguard the technology.
Meanwhile, OpenAI continues to compete fiercely in an increasingly crowded AI agent market, where these agents can handle more complex tasks for users. At a developer event on Tuesday, the company launched a new always-on AI agent called Dots, preparing to compete directly with products such as Muse under Meta Platforms, the parent company of Facebook and Instagram. OpenAI is also adjusting its subscription products, launching a new $500 high-end paid tier that offers higher usage limits and faster processing speeds, while reducing some usage limits for the $200 plan. For most of this year, OpenAI was widely seen as being at a competitive disadvantage, but the company has been firmly consolidating its product layout, focusing on the growth priority of AI agents driven by the most frontier AI large models, and regaining growth momentum in tools that simplify workflows such as programming.
Media previously disclosed in August that OpenAI's annualized revenue accelerated during the summer and exceeded $40 billion. On Tuesday, according to the latest U.S. media reports, a person familiar with the matter said that since July, the company's annualized revenue data, or run-rate revenue, has approached $70 billion, surging by more than 70% since the beginning of the third quarter. That is also why, after this data was disclosed, shares of Oracle, one of the largest builders of OpenAI's AI cloud computing infrastructure, rose sharply. It is worth noting that the run-rate revenue approaching $70 billion is a startup's recent revenue pace annualized, not full-year revenue already achieved in 2026, nor can it all be attributed to ChatGPT as a single product.
AI's "capital buffer compartment": buying time for sustained large-scale expansion of AI compute resources, safety verification, and customer collections
The most fundamental change beneath this unprecedented wave of AI agents is that a single user instruction now triggers an entire workflow that can continue advancing. Meta's Muse runs in a standalone secure virtual machine with a browser, and can continue executing tasks even after the user closes the application. OpenAI's newly launched Dots is powered by the already released GPT-6 Astra, has a cloud computer, and can connect to more than 4,000 applications. Research, programming, procurement, or office tasks can therefore be chained together through planning, retrieval, tool invocation, code execution, and result verification. By extension, compute demand increasingly depends on user scale, task frequency, the inference and execution workload of each task, and peak concurrency demand under enormous AI inference workloads.
The recent advances in the most frontier AI agents and AI large models represented by Muse, Astra, and Anthropic Claude can be said to provide an important technological foundation for the large-scale commercialization expansion of AI applications across industries and the continued surge in AI compute demand. In particular, the expanding scope of agent applications is expected to simultaneously increase demand for AI core infrastructure resources such as AI GPUs and TPUs, high-performance CPUs, high-performance HBM, DRAM, and NAND memory chips for data centers, and high-speed optical interconnect components. From the perspective of AI inference system architecture, GPUs and specialized accelerators handle model computation, CPUs are responsible for turning inference results into actual operations, and memory and storage are responsible for saving and retrieving task states. Long context and multi-turn invocation increase prefill computation and KV cache demand; browsers, code sandboxes, retrieval, and task orchestration increase server CPU load; and file, database, persistent memory, and cache layering extend demand to server DRAM, enterprise SSDs, and high-speed networking. Nvidia's technical materials have already described AI agent inference as an enormous系统工程 spanning GPU HBM, CPU DRAM, local NVMe and remote storage, as well as the internal high-speed optical interconnects that are critical for data transmission.
Safety requirements are adding another layer of real workload. OpenAI has explicitly disclosed that it has added misalignment behavior monitoring to the tool-calling inference for externally deployed Astra, and bears significant computing costs for this. Before Dots performs actions such as sending emails or modifying files, it also passes through an independent automatic review system. This means that, in production environments, the competitive metric is shifting toward "how many tasks can be safely completed per unit cost," with model inference, tool execution, and safety verification jointly determining commercialization efficiency.
From a capital and investment perspective, Anthropic's desire to claim the global IPO throne with a $2 trillion valuation and an IPO fundraising scale of about $100 billion, as well as OpenAI's latest funding round and $1.4 trillion valuation, are beginning to be understood by the market as the "capital buffer compartment" built by Anthropic and OpenAI for the expansion of frontier AI agents. Compute procurement, research and development, and safety investment often occur first, while customer usage and collections are realized gradually afterward. In particular, OpenAI's IPO timing may be adjusted again, so private capital therefore plays the role of spanning these time gaps. Of Anthropic's at least $518 billion in infrastructure commitments over the next decade, about 80% are non-cancelable or must be paid regardless of usage, offering an even more intuitive illustration of the importance of such funding arrangements.