Breaking: Anthropic IPO: User Suspensions and Royalty Claims Ahead of Listing

Anthropic, the developer of the Claude models, is preparing for an initial public offering that it projects at a valuation between 2.5 trillion dollars and 3 trillion dollars. This valuation stands in stark contrast to the company’s current financial reality, which includes a deficit of about 500 billion dollars and annual revenue of less than 20 billion dollars. The company also carries compute commitments to partners like Amazon and Google that reach up to 900 billion dollars. This gap between the projected worth and the actual earnings has become a central point of discussion for investors and users alike.

The company’s path to profitability relies heavily on controlling access to its models and managing the costs of the hardware required to run them. A single Nvidia RTX Pro 6000 graphics card can support about 15 to 20 light users or 1 to 2 heavy coding users under constant use. While heavy users pay 200 dollars a month and light users pay about 20 dollars, the hardware costs remain substantial. The card retails for 20,000 dollars, a price that does not account for the servers, electricity, or data center facilities needed to keep the system running.

For the average consumer, the economics of using these tools are shifting. Most light users do not utilize the chat for more than a few hours a day, which can stretch the bandwidth usage by a factor of ten. However, the acquisition cost for the necessary compute power remains a significant hurdle for the companies providing the service. Nvidia has emerged as a key player in this ecosystem, driving demand while managing supply. The company has allocated most of its available resources to large players like Anthropic and OpenAI, leaving the consumer market to adapt to the resulting costs.

In a significant move to consolidate control, Nvidia agreed to acquire Hugging Face for 12.9 billion dollars. Hugging Face is the primary platform for distributing open AI models and generates about 150 to 200 million dollars in annual revenue during peak years. This acquisition price represents about 70 times the platform’s revenue, a sum that is manageable for Nvidia and its CEO, Jensen Huang. The stated reason for the deal was CUDA, the software that runs these models, but the deeper objective was to control the distribution of models, particularly in the open source market.

The open source sector has been flooded with advanced models from China, including Kimi from Moonshot AI, DeepSeek, and Qwen from Alibaba. These companies have been willing to give their models to the public to foster innovation. Kimi K3 began to outperform Anthropic’s Opus 5 models and was made available to the public. Users on the internet took these large language models, quantized them, and made them usable on smaller machines like Mac Studios and various Nvidia cards. After the acquisition, Hugging Face deleted advanced models from its distribution, removing any model over a quant 4. The 4 bit, 6 bit, 8 bit, and 16 bit full sized models were banned from the platform.

Developers rallied on Discord to stop this perceived corporate overreach and opened Pirate Face, a new location for uploading and downloading models. This effort came after the large 400 gigabyte models at quant 3 and higher were scrubbed from the network. Meanwhile, Anthropic has been engaging with Congress to shape the regulatory environment. The company demanded a ban on Chinese models citing national security, a move that also served to protect its market position. A publicity stunt involving media reports on AI taking over the world, timed with the departure of employee Jacob Coxon, helped drive this agenda.

Congress passed a bill that now bans any government and its affiliates from using any Chinese models. Anthropic is also pushing for bills that would stop other companies, such as Verizon, from launching their own in-house AI models using advanced open source tools. Earlier in the year, the company released Fable 5 and Mythos 5 on June 9, 2026. The U.S. Department of Commerce issued an export control directive that forced Anthropic to suspend access to these models worldwide on June 12, 2026. The models were solving too many problems, leading Anthropic to pull them off the shelf in a move focused on controlling the AI race rather than just winning it.

In recent months, Anthropic has been suspending users with reasons such as abuse, potential fraud, and usage violations. The appeal process has been described as useless because it is fully run by bots, and once an account is kicked off, it is rarely reactivated. With the IPO approaching, these suspensions are expected to multiply 100 times or 1000 times. The company will have deep pockets and access to unlimited funds, which means the end user will likely become less important to its strategy. Anthropic does not make significant money from daily users, nor does it profit heavily from its most active users.

The company is expected to create programs where all user ideas and data will belong to Anthropic. If the company sees great ideas, it gets notified and may suspend users for unproven reasons to secure their intellectual property. This approach allows Anthropic to claim the money and time that users invested in their projects. Dario Amodei, the CEO of Anthropic, stated in an interview that some ideas are worth billions. He suggested that Anthropic will likely have a royalty or partnership program where, if a person or company finds a cure for cancer, some of the accreditation and funding will go to Anthropic.