The Chinese AI startup DeepSeek has recently announced that its AI models have the potential for significant profitability — though with some caveats.
In a tweet on X, DeepSeek proudly claimed that its online offerings boast a “cost profit margin” of 545%. However, this figure is based on “theoretical income.”
In a more detailed analysis shared at the conclusion of a comprehensive GitHub article discussing its strategy for achieving “higher throughput and lower latency,” the company noted that if usage of its V3 and R1 models over a 24-hour timeframe had been charged strictly with R1 pricing, it would have generated an impressive $562,027 in daily revenue.
In contrast, the cost for leasing the required GPUs (graphics processing units) would amount to only $87,072.
However, the firm acknowledged that its actual earnings are “substantially lower” due to various factors, such as nighttime discounts, reduced pricing for V3, and the fact that “only a small selection of services are monetized,” leaving web and app access free of charge.
Naturally, if the app and website were not free, and if other discounts were unavailable, it’s likely usage would decline significantly. Thus, these estimates appear to be quite speculative — more of an indication of potential future profit margins than an accurate portrayal of DeepSeek’s current financial status.
Nevertheless, the company is publicly sharing these figures amidst ongoing discussions regarding the costs and potential profitability of AI. DeepSeek gained significant attention in January after releasing a model that allegedly matched OpenAI’s offerings on certain benchmarks, despite being developed at a markedly lower cost and under U.S. trade restrictions that hinder Chinese entities from acquiring the most advanced chips. This development led to declines in tech stocks and prompted analysts to question AI expenditures.
DeepSeek’s technology impacted more than just Wall Street; its app temporarily overtook OpenAI’s ChatGPT to claim the top spot in Apple’s App Store. However, it has since dropped in the rankings and currently sits at #6 in productivity, trailing behind ChatGPT, Grok, and Google Gemini.
Compiled by Techarena.au.
Fanpage: TechArena.au
Watch more about AI – Artificial Intelligence
