DeepSeek, the new Chinese AI platform, disrupted the tech industry like few others have ever seen before. The availability of the service alone caused huge crashes in the stocks of companies like NVIDIA, which lost up to $600 billion in market capitalization in just one day. However, Mark Zuckerberg is not worried about DeepSeek and confirms that Meta will continue to invest billions in AI development.
DeepSeek ‘scared’ investors who put money into other AI companies
The DeepSeek team revealed that they trained their models with just $6 million. This is just a small fraction of the millions—or billions—that other big names in the niche have invested. Apparently, the company used older NVIDIA H800 AI chips instead of the vendor’s newest hardware. Plus, official benchmarks for the DeepSeek R1 and V3 models show comparable or superior performance to other, more popular ones in some key parameters.
The cost-performance ratio of the DeepSeek models caused a stir on Wall Street. Investors who founded AI companies that spend much more on AI training and development began to have doubts about viability and profitability. NVIDIA is the main supplier of AI chips currently, so it was naturally the most affected. The brand suffered a drop of almost 18% in the price of its shares in 24 hours.
Meta will continue to invest billions in AI, CEO confirms
In the face of this turbulent situation, other big players in the segment have offered their vision on the new and powerful Chinese AI. Among them is Mark Zuckerberg, CEO of Meta, who claims that the company will continue investing billions in its AI models, regardless of the impact of DeepSeek. The firm believes that this will give them a “strategic advantage” in the industry.
“At this point, I would bet that the ability to build out that kind of infrastructure is going to be a major advantage for both the quality of the service and being able to serve the scale that we want to,” Zuckerberg said. He had already revealed last week that the company will spend more than $60 billion on AI throughout this year. Most of the money will be allocated to setting up data centers.
Regarding the potential lower demand for AI hardware due to DeepSeek’s affordable requirements, Zuckerberg said it’s “way too early” to determine that that will actually happen. He noted that AI chips remain crucial for inference purposes.
The controversies that begin to surround DeepSeek
A couple of days ago, OpenAI CEO Sam Altman said that it is “legit invigorating to have a new competitor” to DeepSeek. The company even promised that it will release “much better models.” Later, OpenAI began to suspect potential illegal use of GPT models for training DeepSeek’s. Experts and US officials have also warned of potential risks to user privacy and national security.
Meta took the AI path after the failed venture into the metaverse. What seemed like the future of the tech industry at the time ended up collapsing like a house of cards. However, it seems that the venture into the AI niche is turning out much better for it. Currently, Meta is working on Llama 4, its next-gen LLM. “Our goal with Llama 3 was to make open source competitive with closed models,” said the Meta CEO. “And our goal for Llama 4 is to lead,” he added.
