We've established a new company known as DeepSeek particularly for this purpose. This good friend later founded a company worth tons of of billions of dollars, named DJI. However, LLMs heavily rely upon computational power, algorithms, and data, requiring an initial investment of $50 million and tens of thousands and thousands of dollars per training session, making it troublesome for firms not value billions to maintain. There's still, now it is a whole bunch of billions of dollars that China's placing into the semiconductor business. Within the quantitative subject, High-Flyer is a "high fund" that has reached a scale of a whole lot of billions. Many startups have begun to adjust their methods or even consider withdrawing after main players entered the sector, but this quantitative fund is forging ahead alone. 36Kr: Many consider that for startups, coming into the field after main corporations have established a consensus is not a very good timing. This implies, by way of computational energy alone, High-Flyer had secured its ticket to develop something like ChatGPT earlier than many main tech corporations. Meta isn’t alone - other tech giants are additionally scrambling to understand how this Chinese startup has achieved such outcomes. Meta is worried DeepSeek outperforms its but-to-be-released Llama 4, The knowledge reported.
"We are aware of and reviewing indications that DeepSeek r1 could have inappropriately distilled our fashions, and will share data as we know more," an OpenAI spokesperson mentioned in a remark to CNN. In January 2025, a report highlighted that a DeepSeek database had been left exposed, revealing over 1,000,000 strains of delicate info. When the shortage of excessive-performance GPU chips amongst domestic cloud providers grew to become essentially the most direct factor limiting the start of China's generative AI, in response to "Caijing Eleven People (a Chinese media outlet)," there are no more than five companies in China with over 10,000 GPUs. China-focused podcast and media platform ChinaTalk has already translated one interview with Liang after DeepSeek-V2 was released in 2024 (kudos to Jordan!) On this submit, I translated another from May 2023, shortly after the DeepSeek’s founding. OpenAI, ByteDance, Alibaba, Zhipu AI, and Moonshot AI are among the groups actively finding out DeepSeek, Chinese media outlet TMTPost reported.
Nearly 20 months later, it’s fascinating to revisit Liang’s early views, which may hold the secret behind how DeepSeek, despite restricted assets and compute access, has risen to face shoulder-to-shoulder with the world’s leading AI firms. In reality, this firm, hardly ever considered by the lens of AI, has long been a hidden AI large: in 2019, High-Flyer Quant established an AI firm, with its self-developed deep studying coaching platform "Firefly One" totaling almost 200 million yuan in funding, equipped with 1,a hundred GPUs; two years later, "Firefly Two" elevated its investment to 1 billion yuan, equipped with about 10,000 NVIDIA A100 graphics playing cards. General AI is likely to be one in all the following huge challenges, so for us, it is a matter of easy methods to do it, not why. Despite these challenges, High-Flyer remains optimistic. Within the swarm of LLM battles, High-Flyer stands out as probably the most unconventional player. 36Kr: Are you planning to train a LLM yourselves, or give attention to a specific vertical business-like finance-associated LLMs? Since the release of its latest LLM DeepSeek-V3 and reasoning mannequin DeepSeek-R1, the tech group has been abuzz with pleasure. Besides several main tech giants, this listing includes a quantitative fund firm named High-Flyer. Many of the core members at High-Flyer come from an AI background.
DeepSeek CEO Liang Wenfeng, also the founding father of High-Flyer - a Chinese quantitative fund and DeepSeek’s main backer - just lately met with Chinese Premier Li Qiang, the place he highlighted the challenges Chinese corporations face as a result of U.S. After more than a decade of entrepreneurship, this is the first public interview for this rarely seen "tech geek" type of founder. Therefore, past the inevitable matters of cash, talent, and computational power involved in LLMs, we also mentioned with High-Flyer founder Liang about what sort of organizational structure can foster innovation and how lengthy human madness can final. It bypasses security measures by embedding unsafe matters among benign ones inside a positive narrative. In May, High-Flyer named its new impartial organization devoted to LLMs "DeepSeek," emphasizing its deal with attaining actually human-degree AI. Liang Wenfeng: We can't prematurely design functions based on fashions; we'll deal with the LLMs themselves. Liang Wenfeng: Our venture into LLMs is not straight related to quantitative finance or finance generally. The more essential secret, perhaps, comes from High-Flyer's founder, Liang Wenfeng.