Nvidia’s cutting-edge GPUs, essential for training state-of-the-art AI models, are currently in such high demand that it feels almost unquenchable. Their data center division alone pulled in over $30 billion in the third quarter—a staggering increase, nearly tenfold, from just two years back. Tech behemoths are in a mad rush, shelling out unimaginable sums to erect AI data centers powered by Nvidia GPUs.
The AI frenzy has seen each new model outdo the previous one. Take OpenAI’s GPT-4, which far surpasses GPT-3, or Alphabet’s cutting-edge Gemini models, which outperform their predecessors significantly. These leaps forward, however, come with hefty price tags.
Start Your Mornings Smarter! Get Breakfast news delivered directly to your inbox every market day. Sign Up For Free »
The cost to train GPT-4 is estimated at a whopping $100 million, compared to just millions for GPT-3. Anthropic’s CEO, Dario Amodei, projects the next generation could soar to a billion dollars in production costs. Investing in thousands of high-performance GPUs is expensive, and collecting vast amounts of training data is a challenge of its own.
AI models like GPT-4 operate by predicting upcoming tokens in a sequence, a method that’s generally effective. The pinnacle AI models churn out superb text, forge captivating images, and occasionally demonstrate advanced reasoning. More data typically leads to better models, as does extended time spent processing this data.
However, we’ve run up against a wall. Even with more data and more computing power, the pace of improvement in AI models is slowing. Venture capitalist Marc Andreessen has pointed out how AI models appear to be capping out, regardless of data or hardware enhancements.
Breaking through this limit could revolutionize the field, but there’s a chance large language models (LLMs) might not be capable of significantly more than what we see now. The surging demand for AI chips hinges on the belief that investing $1 billion to $10 billion in AI models is worth it. But what if it’s not?
If we’ve hit the ceiling on AI capabilities, the rash of billion-dollar investments by tech companies might not deliver the anticipated returns, potentially leading to harsh consequences for companies like Nvidia if the demand for AI chips nosedives.
Nvidia has offered jaw-dropping returns for investors, firmly holding its position as a leader in the AI chip market. Yet, the reality remains: no trend grows indefinitely. An AI breakthrough could bypass current limitations, but it’s equally possible that AI, like most groundbreaking technologies in history, is experiencing some level of hype.
Even if AI stops evolving rapidly, artificial intelligence isn’t disappearing anytime soon. Still, Nvidia’s phenomenal growth and profits may decline if their models’ advancement slows. Considering Nvidia’s current market valuation hovering around $3 trillion, its stock might seem precarious.
Ever get the feeling you’ve missed the train on buying top-performing stocks? There’s something you should know.
Occasionally, our team of experts issues a “Double Down” recommendation for companies they believe are poised for impressive growth. If you fear missing your opportunity, right now is an opportune moment to invest before it’s too late. Here’s the record:
– Nvidia: Investing $1,000 when we doubled down in 2009 would now be worth $355,011!*
– Apple: Putting $1,000 into the stock when we doubled down in 2008 would have grown to $44,516!*
– Netflix: A $1,000 investment from our 2004 double down would now be at $470,586!*
Today, we’re sounding the “Double Down” alarm for three outstanding companies, and this chance might not come around again soon.
Discover 3 “Double Down” Stocks »
*Stock Advisor returns as of November 25, 2024
Suzanne Frey from Alphabet serves on The Motley Fool’s board. Timothy Green doesn’t hold positions in the mentioned stocks. The Motley Fool has stakes in and endorses Alphabet and Nvidia. Our disclosure policy »
This article, “Prediction: This Massive Risk Could Derail Nvidia Stock,” was originally published by The Motley Fool.
