DeepSeek, a Chinese artificial intelligence startup, recently made headlines by announcing that its AI models could be quite profitable, but with some important caveats. The company took to the social media platform X to share its impressive figures, claiming a staggering “cost profit margin” of 545%. However, it’s crucial to note that this figure is based on what they call “theoretical income,” which means it’s not necessarily reflective of the actual money they are making right now.
In a detailed post on GitHub, DeepSeek elaborated on how they arrived at these numbers. They explained that when analyzing the usage of their AI models, specifically the V3 and R1 versions, over a 24-hour period, they estimated that if all usage had been charged at the R1 pricing, they could have generated a whopping $562,027 in daily revenue. This sounds impressive at first glance, but the reality is a bit more complicated. The cost of leasing the powerful graphics processing units (GPUs) necessary for running these models would have only amounted to $87,072.
Despite these eye-catching figures, the company acknowledged that its actual revenue is “substantially lower” than these projections. Several factors contribute to this discrepancy, including discounted rates during nighttime hours, lower pricing for the V3 model, and the fact that many of their services are not monetized yet. For instance, access to their web and app platforms remains free for users, which means they are not currently charging for a significant portion of their offerings.
This raises an interesting point: if DeepSeek decided to charge for their app and website, or if they eliminated discounts, it’s likely that the overall usage would decrease. This suggests that their calculations may be more speculative than concrete, serving more as a glimpse into potential future profits rather than an accurate representation of their current financial status.
The timing of DeepSeek’s announcement is particularly noteworthy, as it coincides with ongoing discussions about the costs and profitability of AI technologies in the industry. The company gained attention back in January when it introduced a new model that reportedly matched the performance of OpenAI’s model, known as o1, on certain benchmarks. What’s more impressive is that DeepSeek achieved this feat at a significantly lower cost, all while facing U.S. trade restrictions that limit access to the most advanced chips for Chinese companies.
DeepSeek’s advancements didn’t just stir the pot on Wall Street; they also made waves in the consumer market. At one point, their app even surpassed OpenAI’s ChatGPT to claim the top spot on Apple’s App Store. However, it has since dropped in rankings and is currently positioned at No. 6 in the productivity category, trailing behind ChatGPT, Grok, and Google Gemini.
In summary, while DeepSeek’s claims of high profit margins and impressive revenue projections are certainly eye-catching, they come with significant caveats. The company’s actual financial performance is likely much lower due to various discounts and the free access they provide to many services. As the AI landscape continues to evolve, it will be interesting to see how DeepSeek navigates these challenges and whether it can turn its theoretical profits into real-world success. The ongoing debates about AI profitability will undoubtedly shape the future of companies like DeepSeek as they strive to carve out a sustainable niche in this competitive market.
