On one side are the skeptics and shorts, betting that Nvidia stock will soon see a brutal correction like other tech companies in the past.
On the other are the enthusiasts and the purchased, who believe that Nvidia’s climb to the top is just beginning. Some even project that it could surpass Apple in market value.
Until now, the short sellers are bleeding. The loss last Thursday alone, with the marking to market of short positions, was US$ 826 million; in the year, exceeds US$ 11 billion.
In January, the market cap from Nvidia was “only” $370 billion. In seven months, that figure more than tripled, as the world began to understand the impact of artificial intelligence and Nvidia’s role in the game.
The company is now worth $1.14 trillion, almost like Amazon ($1.37 trillion) and not far from Alphabet ($1.64 trillion). Apple remains far at the top, with $2.8 trillion market cap.
According to Barron’s, the average price target for Nvidia stock among analysts surveyed by FactSet is $574 – one upside 25% potential that would put your market cap over $1.4 trillion.
One of the most aggressive analysts on the company’s potential is Chaim Siegel of Elazar Advisors, a tech research firm that works for hedge funds and family offices.
Siegel has a price target of $1,606, which would make Nvidia worth $4 trillion, something no other company has ever achieved.
Nvidia has racked up staggering sales and profits because of its absolute dominance in supplying the chips used in generative artificial intelligence systems like ChatGPT and Bard.
Revenue at its data center division, which includes processor sales, hit $10.3 billion in the second quarter, more than double the previous quarter. It was the highlight in the presentation of results last week, when the Californian company reported revenues of US$ 13.5 billion.
After Nvidia beat the consensus (by far), the vast majority of analysts raised the stock’s price target. The company currently has the best hardware and software solutions to equip artificial intelligence developers with large-scale language models (LLMs), systems capable of interpreting data and creating original content from them.
“A new era of computing has begun,” Nvidia CEO and co-founder Jensen Huang said at Wednesday’s quarterly earnings call. “Companies around the world are transitioning from general purpose computing to accelerated computing and generative artificial intelligence.”
Nvidia has been preparing for this moment for a decade and maintains, at least until now, an isolated lead in this new market.
A milestone in the company’s history was the launch of its innovative GeForce 256 chip in 1999, seen as the first graphics processing unit (GPU). The processor made a big impact on image rendering. Later, GPUs were reconfigured to run artificial intelligence systems. Its architecture allows carrying out activities in parallel, accelerating computational capacity.
Analysts at Goldman Sachs say that while there are some emerging competitors, Nvidia should “maintain its status as a standard in the fast-paced computing industry for the foreseeable future, given its competitive edge and the urgency with which customers are developing and delivering complex AI models.”
Goldman raised its price target for the share from US$495 to US$605. Itau BBA went from US$500 to US$600.
For Thiago Kapulskis, head of global tech analysis at Itaú, traditional processors do not have enough computational power to meet the demands of AI.
Nvidia GPUs have been the best answer so far, but tapping into their potential depends on the software required for model training – and then CUDA (Compute Unified Device Architecture), a software development and training platform, is another Nvidia’s differential, interconnecting programmers and developers.
Nvidia is at the heart of the AI transformation, say Kapulskis and the Itaú team, with a wide range of services and products. “This gives us confidence to raise estimates not just for this year, but beyond 2024.”
Kapulskis estimates that data center revenue could reach $78 billion by 2025.
Bank of America has adjusted its price target from $450 to $550, saying it sees Nvidia as too hard to match the competition. The biggest risk right now, says BofA, are possible sales restrictions in the Chinese market.
Other analysts see potential pitfalls ahead for Nvidia’s scale and the spread of artificial intelligence.
Firstly, there is a bottleneck in accessing processors, especially the latest generation GPUs, the H100. ChatGPT uses about 10,000 of these boards to run their models.
Nvidia and TSMC – which handles the physical production of processors – are working to resolve the bottleneck, but access remains restricted. When there is a balance between supply and demand, artificial intelligence capabilities should advance more quickly, because today many startups and smaller companies are unable to buy the H100, whose list price is US$ 10,000.
Another possible bottleneck for the advancement of AI is the availability of electricity.
According to the International Energy Agency, data centers and data transmission systems make up somewhere between 2% and 3% of all electricity demand – and the number is expected to grow at the pace of AI expansion.
According to Bloomberg, the processing of Google’s AI systems alone uses enough electricity for residential consumption in Atlanta, a city of 500,000 inhabitants.
None of this should stop the relentless march of technological innovation – and generative AI is the revolution of the moment.
Beth Kindig, CEO and tech analyst at manager I/O Fund, was one of the first to foresee Nvidia’s potential for ascension five years ago due to advances in AI.
Until then, the company was recognized only for the quality of its boards in rendering images for games and animations due to the development of artificial intelligence.
Beth has covered the company since 2018. That year, she published a report saying that “Nvidia is already the universal platform for artificial intelligence development, but that won’t become obvious until this innovation matures.”
Developers, she said, were coding future AI applications with Nvidia because it offered better, simpler, and more flexible solutions than other vendors.
“When artificial intelligence matures, you can expect data center revenue to be the top segment in Nvidia’s revenue,” he wrote.
Years later, in an article published in Forbes in August 2021, Beth returned to the charge: “Here is the reason why the valuation Nvidia will surpass Apple in 5 years.”
In Beth’s logic, Nvidia would be close to winning “with the economy of artificial intelligence” which, according to numbers presented by her, will be a much larger market than the economy of the mobile internet.
On that date, Nvidia was worth US$ 550 billion on the Stock Exchange, against US$ 2.5 trillion for Apple.
For Beth, the short sellers err in cataloging Nvidia as a passing bubble of the artificial intelligence boom.
“The company has a virtual monopoly (95% share) in the GPU data center market, which is expected to grow at a 23% CAGR through 2030,” she told Fox Business this week.
Giuliano Guandalini