DeepSeek’s AI Model Shakes Global Tech Markets, Triggering Nvidia Stock Plunge and US Concerns

China’s DeepSeek has shaken the global AI industry with claims of developing a highly advanced model at a fraction of the cost incurred by US tech giants. However, skeptics are questioning its legitimacy.

DeepSeek’s AI Model Shakes Global Tech Markets, Triggering Nvidia Stock Plunge and US Concerns

The AI industry is witnessing a significant upheaval as DeepSeek, a Hangzhou-based startup, asserts that it has developed an advanced AI model at a fraction of the cost spent by US tech leaders like OpenAI and Google. This claim, if true, challenges the dominance of American AI companies, which have historically commanded billion-dollar investments to build cutting-edge models.

The revelation that DeepSeek trained its AI model R1 using just $5.6 million and 2,000 Nvidia H800 GPUs—a chip designed to comply with US export restrictions—has raised eyebrows across the industry. Comparatively, OpenAI’s GPT-4 is estimated to have cost over $100 million to train, using 25,000 high-end H100 GPUs.

The AI community is now engaged in intense scrutiny of DeepSeek’s claims, with some experts dismissing them as exaggerated or misleading. Meanwhile, Nvidia’s stock price plummeted 17%, erasing nearly $593 billion in market value, before partially recovering the next day.

DeepSeek’s Announcement and Its Claims

  • DeepSeek was founded in late 2023 by Liang Wenfeng, a serial entrepreneur in the AI industry.
  • The company unveiled DeepSeek V3, a 671 billion-parameter model, and DeepSeek R1, an advanced reasoning model.
  • DeepSeek claims it used 2,000 Nvidia H800 GPUs and spent $5.6 million to train R1’s foundational model, V3.
  • This sharply contrasts with OpenAI’s GPT-4, which reportedly cost over $100 million to train.
  • The startup’s success has raised concerns about the effectiveness of US chip export restrictions aimed at limiting China’s AI advancements.

Skepticism from AI Experts and Industry Leaders

  • Pedro Domingos, a professor emeritus at the University of Washington, expressed doubt about DeepSeek’s claims:
    • “It’s very much an open question whether DeepSeek’s claims can be taken at face value. The AI community will be digging into them and we’ll find out.”
    • He acknowledged that training a model with $6 million is plausible but added that it could be merely fine-tuning an already expensive model.
  • Palmer Luckey, founder of Oculus VR, outright dismissed DeepSeek’s claims, calling them “bogus”.
    • He accused the startup of engaging in propaganda to mislead American investors, stating:
    • “It is pushed by a Chinese hedge fund to slow investment in American AI startups, service their own shorts against American titans like Nvidia, and hide sanction evasion.”
  • Alexandr Wang, CEO of Scale AI, suggested that DeepSeek had access to 50,000 more advanced H100 GPUs, but he did not provide evidence.
  • Elon Musk, a close ally of US President Donald Trump, also voiced skepticism, responding “Obviously” on X to a post questioning DeepSeek’s claims.

Financial and Market Impact

  • DeepSeek’s announcement caused a sharp decline in Nvidia’s stock price, which dropped 17% on Monday, wiping out nearly $593 billion in market value—equivalent to Sweden’s GDP.
  • However, on Tuesday, Nvidia’s stock rebounded by nearly 9%, as initial concerns began to fade.
  • The Nasdaq 100 index, which is heavily weighted toward tech stocks, also saw a 1.59% recovery after its steep decline.

China’s AI Progress and US Export Controls

  • DeepSeek’s rapid progress in AI has cast doubt on the effectiveness of US chip export restrictions.
  • Zihan Wang, a PhD candidate who worked on an earlier DeepSeek model, dismissed skepticism, stating:
    • “Talk is cheap. If they’d spend more time working on the code and reproduce the DeepSeek idea themselves, it would be better than talking on the paper.”
  • In a past interview, Liang Wenfeng revealed that DeepSeek had stockpiled 10,000 Nvidia A100 chips before US export bans took effect.

Limitations and Censorship in DeepSeek’s Model

  • Despite its technological advancements, DeepSeek’s AI model R1 reportedly censors topics that are politically sensitive in China, including:
    • The 1989 Tiananmen Square massacre.
    • The status of Taiwan.

Experts’ Analysis on DeepSeek’s Cost Efficiency

  • Tim Miller, an AI professor at the University of Queensland, noted the uncertainty around DeepSeek’s claims:
    • “The model itself gives away a few details of how it works, but the costs of the main changes they claim don’t ‘show up’ in the model itself so much.”
    • He pointed out that while breakthroughs do occur, this development seemed “too good to be true.”
  • Lucas Hansen, co-founder of CivAI, provided insights into why DeepSeek’s model may have been cheaper to train:
    • “There have been a lot of algorithmic and hardware improvements since 2022, driving down the cost of training a GPT-4 class model.”
    • He added that the real breakthrough was not the base model but the fine-tuning process, which is comparatively inexpensive.

Source: Al Jazeera and NDTV

Supreme Court Dismisses Compassionate Appointment Appeal, Calls for Hand-to-Mouth Conditions
Legal Wires
Supreme Court Dismisses Compassionate Appointment Appeal, Calls for Hand-to-Mouth Conditions
Supreme Court flags conflicting judgments on whether compassionate appointment claims should be based on the scheme prevailing at the time of death or when the application is considered.
Kerala High Court Upholds Constitutionality of Brain Death Certification, Dismisses PIL Against THOTA Provisions
Legal Wires
Kerala High Court Upholds Constitutionality of Brain Death Certification, Dismisses PIL Against THOTA Provisions
The Kerala High Court dismisses PIL challenging brain death certification and upholds the constitutional validity of THOTA, emphasizing Parliament’s authority to define medical standards for brain death.
Supreme Court Upholds Father’s Custody Rights, Overrules High Court’s Ruling in Child Custody Battle
Legal Wires
Supreme Court Upholds Father’s Custody Rights, Overrules High Court’s Ruling in Child Custody Battle
The Supreme Court ruled that a father’s claim to custody is superior to that of his child’s maternal grandparents, emphasizing the father as the natural guardian, overturning a High Court verdict.
Or
Powered by Lit Law
New Chat
Sources
No Sources Available
Ask AI