Nvidia will definitely be disrupted

Latest update time：2024-11-10

Reads：

????If you hope to meet more often, please mark the star ?????? and add it to your collection~

Source: The content comes from Semiconductor Industry Observer, thank you.

Tech giant Nvidia Corp.’s seemingly impenetrable moat in artificial intelligence has fueled a surge in its stock price. But attention is growing on potential disruptors, particularly startups focused on emerging AI challenges where Nvidia may be at a disadvantage.

These upstarts are focusing on key questions. Is it worth investing massively in AI data centers powered by Nvidia chips? And is it reasonable for the tech industry to rely on just one chip maker to get its hands on the hottest trends in tech?

Cerebras, an AI chip startup that just filed to go public, vowed to “accelerate the development of AI, making it faster and easier to use” and “popular around the world.” Rodrigo Liang, CEO of SambaNova, another Nvidia challenger, also mentioned the need to expand the scope of AI applications, pointing to growing concerns about “skyrocketing” electricity costs and diminishing returns.

“You have to get costs down to get performance up to the point where developers can innovate and do it at scale,” he told Investor’s Business Daily. “If you don’t do that, then it’s just a game that only the big players can play.”

“We are the alternative to Nvidia,” Liang declared.

Nvidia, Dominating the Training Market

Becoming an alternative to Nvidia has been the goal of many startups since chip giant Nvidia began to dominate the field of artificial intelligence with the rise of deep learning a decade ago.

That was something that was unexpected.

Deep learning, or building artificial neural networks to mimic the way human memory works, requires a lot of processing power. Nvidia’s graphics processors, used in video games and Hollywood blockbusters, have proven to be good at this job.

“Deep learning played into their hands,” Naveen Rao, who co-founded AI chip startup Nervana in 2014 and is now an executive at DataBricks, said in a 2019 interview. Nvidia’s chips have proven particularly well suited to a key step in deep learning, which is training AI models by using big computations to crunch reams of data.

Training AI models "requires a lot of data and a lot of processing," IDC analyst Shane Rau told IBD. "You throw a million pictures of cats at the model and ask it to 'recognize a cat.'"

In the next deep learning phase, the inference phase, the model performs actions based on the knowledge developed in the model. “You flip the model over and ask the model to recognize a cat, and you feed it a picture of a cat very quickly,” Rau said.

The industry refers to inference as “AI in production” or “AI in action.”

"Training is like going to college, and reasoning is like going to work," Tony Kim, managing director at SambaNova investor BlackRock, told IBD.

Inference is expected to become an increasingly large need in the AI space. Startups see a huge opportunity. The move to inference requires more custom chips for specific tasks. Those processors are expected to use less power and be less expensive. "As models start to stabilize ... more people will invest less in training and do inference," Lowe said.

The push for lower-cost AI computing could lead to more competition.

“Not everyone is going to spend $30,000 on a GPU,” Rau said. “If you can run a more specialized AI model on a chip that’s cheaper, that’s a good goal.”

Reasoning, there are better options

Liang said that’s the goal of SambaNova, which the company’s website touts as “the world’s fastest AI inference.”

The Palo Alto, California-based company has raised more than $1.1 billion in venture capital from investors including BlackRock, GV, Intel Capital and SoftBank Vision Fund. Its major clients include Saudi Aramco, Lawrence Livermore Laboratory and Argonne National Laboratory. The 600-employee startup was valued at more than $5 billion in 2021.

Liang echoed that sentiment, saying that while Nvidia’s powerful and power-hungry chips are great for training, inference requires a different type of processor. “You can’t continue to use brute force to solve the problem,” he said, advocating for “finding smarter and more efficient ways to achieve scale.” “You need a different kind of chip,” Liang said. “That’s a huge market for us.”

BlackRock’s Kim sees the same opportunity, saying: “Inference AI running in production is going to grow like crazy. These startups are finding life here. This inference demand is starting to explode.”

BlackRock’s policy is not to comment on specific companies. Kim offered his own personal take on the players in the AI market. He said companies looking to adopt AI face a key dilemma. “The investment required is so huge,” which “leads to greater dependence on a single vendor.”

Kim said there’s no doubt that Nvidia’s technology is impressive. But its dominance has prompted “a whole slew of other companies to say, ‘Hey, we need to, we need to build alternatives. Can we rely 100% on one company?’” “No company wants to be 100% dependent on one company,” he said.

The potential size and significance of the AI inference market remains unclear.

Bernstein analysts speculated in a client note this month that the number will be "very large," but it all depends on the return on investment of future AI models. "We don't have an answer for this - while we generally believe it, it's hard to be convincing anyway," Bernstein analysts wrote.

Meanwhile, Nvidia isn’t standing still. “Over the last four quarters, we estimate that inference has contributed more than 40% of Nvidia’s data center revenue,” Chief Financial Officer Colette Kress told analysts in August.

A spokesperson for Nvidia declined to comment further.

Challenges are not easy, success is hard

Nvidia will likely remain a dominant force in AI. But the AI market is still evolving.

“As most models evolve, their market grows, they grow, grow, grow, grow, and eventually they become segmented,” Raw said.

BlackRock's Kim said the training space, currently dominated by Nvidia, is "difficult for anyone to enter." In addition, other chip giants led by AMD and large cloud computing companies such as Microsoft and Google have also joined the competition. "Even if this market is so large, it's still difficult for you to succeed."

However, he said the AI market is huge and is expected to grow "100-fold or 1,000-fold."

“I think some startups can succeed,” Kim added.

That wasn’t the case with Nervana, which was a potential challenger to Nvidia and was acquired by Intel in 2016. Intel eventually shut down Nervana’s operations a few years later after acquiring another AI chip startup, Habana.

But the risk of exploring alternatives to Nvidia is worth it. “There’s room for more than one company called Nvidia,” argues BlackRock’s Kim.

Andrea Schulz, managing partner at financial advisory firm Grant Thornton, said AI chip startups must contend with another serious hurdle: It’s generally more difficult for new companies to enter the market for hardware technology.

"Hardware is hard to do," she told IBD. "That's been said a lot, and it's absolutely true." She said the long-term prospects for AI chip startups remain uncertain. It's unclear "whether they're going to disrupt incumbents and whether they're going to be able to be incumbents in the ecosystem for a long time, or whether they're going to be acquired as they go along," she said. "I think that's the more likely path.

Meanwhile, Liang remains optimistic about the company he co-founded in 2017, which he expects will play a key role in what is expected to be a long battle. “You can’t win overnight,” he said. “You have to persist for a long time. And as we’ve seen, not everyone can persist for a long time.”

In some ways, the AI war is a protracted dance in which players must go with the flow. That’s where companies like SambaNova come in, Liang said.

Liang, who grew up in Brazil, recalled that when deciding on the startup’s name, he “wanted to find a name that was Brazilian.”

His team chose “SambaNova,” which Liang said was a “very, very perfect match” for the neural network they were building.

“Samba is a dance,” he said. “It flows. Data flows. We ended up loving the name.”

Liang also drew lessons from the dot-com era, when his tech career took off and another chip giant seemed invincible. Back then, he said, “no one thought they could disrupt Intel. Intel had more than 90% of the market share. But now it’s different.”

“The best technology wins,” he said. “I think it’s as simple as that: The best technology wins. I don’t think it’s about brand. You have to produce the best technology at scale. People want energy efficiency, cost efficiency and performance. If you get those, the best technology wins. And that’s what we really optimize for.”

NVIDIA, singing all the way

But as many companies boast about subverting Nvidia, we find that Nvidia's market value has once again hit a new high.

Nvidia shares rose to a record high on Thursday, making the chipmaker the first company in history to have a market value of more than $3.6 trillion, as Wall Street's stock rally sparked by Donald Trump's return to the White House continued, Reuters reported. Investors were generally optimistic about tax cuts and deregulation after the Republican candidate won the election on Tuesday, boosting shares of the dominant artificial intelligence chipmaker by 2.2%.

Nvidia's stock market value closed at $3.65 trillion, surpassing Apple. Nvidia's market value reached a record closing price of $3.57 trillion on Oct. 21, and then surpassed the iPhone maker on Tuesday to become the world's most valuable company, according to data from the London Stock Exchange. Apple shares rose 2.1% on Thursday, with a market value of $3.44 trillion.

As Microsoft and Google battle to dominate AI, Nvidia has emerged as the biggest winner in the U.S. stock market. The Silicon Valley chip designer’s shares rose 12% in November, tripling in value by 2024.

Analysts on average expect Nvidia to report a more than 80% rise in quarterly revenue to $32.9 billion when it reports results on Nov. 20, according to the London Stock Exchange. Nvidia briefly became the world's most valuable company in June but has since been surpassed by Microsoft Corp. , opens in a new tab and Apple. The three tech companies have been neck and neck in terms of market value for months. Microsoft, which has a market value of nearly $3.16 trillion, saw its shares rise 1.25% on Thursday.

END

????Semiconductor boutique public account recommendation????

▲Click on the business card above to follow

Focus on more original content in the semiconductor field

▲Click on the business card above to follow

Focus on the trends and developments of the global semiconductor industry

*Disclaimer: This article is originally written by the author. The content of the article is the author's personal opinion. Semiconductor Industry Observer reprints it only to convey a different point of view. It does not mean that Semiconductor Industry Observer agrees or supports this point of view. If you have any objections, please contact Semiconductor Industry Observer.

Today is the 3942nd content shared by "Semiconductor Industry Observer" for you, welcome to follow.