Hock Tan, Broadcom CEO (left) and former Intel CEO Pat Gelsinger.
Reuters | CNBC
It's been a big year for silicon in Silicon Valley, but a tough year for the company most responsible for the region's title.
intel, The 56-year-old chipmaker, co-founded by industry pioneers Gordon Moore, Robert Noyce and legendary investor Arthur Rock, had its worst year since going public in 1971, losing 61% of its value.
The opposite story unfolded in Broadcomthe chip group run by CEO Hock Tan and headquartered in Palo Alto, California, about 15 miles from Intel's Santa Clara campus.
Broadcom's stock price is up 111% in 2024 as of Monday's close, its best performance ever. The current company is the result of the 2015 acquisition of Avago, which went public in 2009.
The driving force behind the differing narratives has been artificial intelligence. Broadcom has jumped on the AI train, while Intel has largely missed it. The shifting fortunes of the two chipmakers underscore the fleeting nature of leadership in the technology industry and how a few key decisions can lead to hundreds of billions — or even trillions — of dollars in market value shifts.
Broadcom develops custom chips for Google And other huge cloud companies. It also makes the essential networking equipment that large server clusters need to connect thousands of AI chips together. In the AI space, Broadcom is largely overshadowed Nvidiawhose graphics processing units, or GPUs, power most of the large language models developed at OpenAI, MicrosoftAnd Google and Amazon And also enabling more massive AI workloads.
Although less well-known, Broadcom's accelerator chips, which the company calls XPUs, have become an essential part of the AI ecosystem.
“The reason it's really up is because they're talking about AI, AI, AI, AI,” Eric Ross, chief investment strategist at Cascend, told CNBC's “Squawk Box” earlier this month.
Intel, which for decades was the dominant chip maker in the United States, has been left out of the AI field. Its server chips lag far behind Nvidia's chips, and the company has also lost market share to its longtime rival Advanced micro devices While spending heavily on new factories.
Intel's board of directors ousted Pat Gelsinger as CEO on December 1, after a tumultuous four-year tenure.
“I think someone more innovative might have seen the AI wave coming,” Paul Argenti, a management professor at Dartmouth’s Tuck School of Business, said in an interview on “Squawk Box” after the announcement.
An Intel spokesman declined to comment.
Broadcom is now worth about $1.1 trillion and is the eighth U.S. technology company to cross the $1 trillion mark. It is the second most valuable chip company, after Nvidia, which has led the AI boom to a valuation of $3.4 trillion, trailing only apple Among all public companies. Nvidia's stock price is up 178% this year, but it actually did better in 2023, when it rose 239%.
Until four years ago, Intel was the world's most valuable chipmaker, with a market value approaching $300 billion in early 2020. The company is now worth about $85 billion, and has just been dropped from the Dow Jones Industrial Average — and solved. It was replaced by Nvidia – which was in talks to sell core parts of its business. Intel now ranks 15th by market capitalization among semiconductor companies globally.
“It's not meant for everyone”
After the Avago-Broadcom merger in 2015, the combined company's largest business was chips for TV set-top boxes and broadband routers. Broadcom still makes the Wi-Fi chips used in laptops as well as the iPhone and other smartphones.
After a failed attempt to buy giant mobile phone chips Qualcomm In 2018, Broadcom turned its attention to software companies. The peak of the spending spree came in 2022 with the announcement of the $61 billion acquisition of server virtualization software company VMware. Software accounted for 41% of Broadcom's $14 billion in revenue last quarter, thanks in part to VMware.
What's exciting on Wall Street is Broadcom's role in working with cloud providers to build custom AI chips. The company's XPUs are generally simpler and less expensive to operate than Nvidia's GPUs, and are designed to run specific AI programs efficiently.
Cloud vendors and other major internet companies spend billions of dollars annually on Nvidia GPUs so they can build their own models and run customers' AI workloads. Broadcom's success with custom chips sets up an AI spending showdown with Nvidia, as large-scale cloud companies look to differentiate their products and services from their rivals.
Broadcom chips are not available to everyone, as only a few companies can afford to design and build their own custom processors.
“It should be Google, it should be Google deadmust be Microsoft or oracle “To be able to use those chips. These chips are not for everyone,” Harsh Kumar, an analyst at Piper Sandler, told CNBC's “Squawk on the Street” on Dec. 13, the day after Broadcom's earnings.
Although 2024 was a breakout year for Broadcom, with AI revenue increasing 220%, December put it into record territory. The stock was up 45% during the month as of Monday's close, 16 percentage points better than the previous best month.
On the company's December 12 earnings call, Tan told investors that Broadcom had doubled its XPU shipments to its three hyperscale providers. The most famous of the bunch is Google, which relies on the technology for its Tensor Processing Units, or TPU, used to train Apple's AI software released this year. The other two clients, according to analysts, are ByteDance and Meta, the parent company of TikTok.
Within about two years, companies could spend $60 billion to $90 billion on XPUs, Tan said.
“In 2027, we believe each of them plans to deploy 1 million XPU clusters across a single fabric,” Tan said of the three hyperscale customers.
In addition to AI chips, AI server clusters need powerful networking hardware to train more advanced models. Dedicated AI networking chips accounted for 76% of Broadcom's $4.5 billion networking sales in the fourth quarter.
Broadcom said a total of about 40% of its $30.1 billion in semiconductor sales in 2024 were AI-related, and that AI revenue would increase 65% in the first quarter to $3.8 billion.
“The degree of success among hyperscalers in their initiatives here is clearly up for debate,” Cantor analyst CJ Muse, who recommends buying Broadcom shares, wrote in a Dec. 18 report. “But any way you slice it, the focus here will still be a meaningful boon for those who benefit from custom silicon.”
Intel had a very bad year
Before 2024, Intel's worst year on the market was 1974, when the stock fell 57%.
The seeds of the company's recent missteps were sown years ago, as Intel failed to buy mobile chips from Qualcomm, ARM, and Apple.
Rival AMD has begun to gain market share in the critical PC and server CPU markets thanks to its productive manufacturing relationship with Taiwan Semiconductor Manufacturing Company. Intel's manufacturing process has been lagging behind for years, resulting in slower and less energy-efficient central processing units, or central processing units (CPUs).
But Intel's most costly whiff is in AI, which is a big reason to fire Gelsinger.
Originally created for video games, Nvidia's GPUs have become important devices in developing power-hungry AI models. Intel's CPU, once the most important and expensive part of a server, has become an afterthought in an AI server. The GPUs that Nvidia will ship in 2025 don't even need an Intel CPU — many of them are paired with an ARM-based chip designed by Nvidia.
Since Nvidia has reported revenue growth of at least 94% over the past six quarters, Intel has been forced to scale back. Sales have declined in nine of the past 11 periods. Intel announced last August that it would eliminate 15,000 jobs, or about 15% of its workforce.
“We are working to create a leaner, simpler, more agile Intel,” Chairman Frank Geary said in a Dec. 2 press release announcing Gelsinger's departure.
One of the big problems Intel faces is its lack of a comprehensive AI strategy. It has touted investors for the AI capabilities in its laptop chips, and released a competitor to Nvidia called the Gaudi 3. But neither the company's AI PC initiative nor its Gaudi chips have gained much traction in the market. Intel Gaudi 3 sales missed the company's $500 million goal for the year.
Late next year, Intel will launch a new AI chip codenamed Falcon Shores. It will not be built on the Gaudi 3 architecture, but will instead be a GPU.
“Will it be great? No, but it's a good first step in getting the platform done,” Michelle Holthaus, Intel's interim co-CEO, said at a Barclays financial conference on Dec. 12.
Holthaus and fellow interim CEO David Zinsner have pledged to focus on Intel products, leaving the fate of Intel's expensive foundry division unclear.
Before his departure, Gelsinger championed a strategy that included Intel finding its niche in the semiconductor and chip manufacturing market to compete with TSMC. In June, at a conference in Taipei, Gelsinger told CNBC that when its factories start operating, Intel wanted to build “AI chips for everyone,” giving companies like Nvidia and Broadcom an alternative to TSMC.
Intel said in September that it plans to spin off its foundry business into an independent unit with its own board and the ability to raise outside capital. But right now, Intel's primary customer is Intel. The company said it does not expect significant sales from outside customers until 2027.
At a Barclays event this month, Zinsner said a separate board of directors for the foundry business “is taking action today.” More broadly, he noted that the company is looking to remove complexity and associated costs wherever possible.
“We will constantly scrutinize where we are spending money, to make sure we are getting the right return,” Zinsner said.
Watch: Intel plans to take chip company Altera public
