Broadcom CEO Hock Tan (left) and former Intel CEO Pat Gelsinger.
Reuters | CNBC
It’s been a big year for Silicon Valley’s silicon industry, but it’s also been a brutal year for the companies most responsible for the region’s nickname.
Intel, The 56-year-old chipmaker, co-founded by industry pioneers Gordon Moore and Robert Noyce and legendary investor Arthur Rock, has gone through its own twist this year. It was the worst year since it went public in 1971, with its market value shrinking by 61%.
The opposite story happened in BroadcomIt is a chip group run by CEO Hock Tan and is headquartered in Palo Alto, California, about 15 miles from Intel’s Santa Clara campus.
As of Monday’s close, Broadcom’s stock price has soared 111% in 2024, its best performance ever. The current company is a product of a company Acquisition in 2015 Developed by Avago and launched in 2009.
The driving force behind the different narratives is artificial intelligence. Broadcom jumped on the artificial intelligence train, while Intel basically missed the boat. The changing fortunes of the two chipmakers underscore the ephemerality of technology industry leadership and how a few key decisions can result in changes in market capitalization of hundreds of billions or even trillions of dollars.
Broadcom develops custom chips Google and other large cloud companies. It also provides the critical networking equipment needed for large server clusters to connect thousands of artificial intelligence chips together. In the field of artificial intelligence, Broadcom’s brilliance has been largely NVIDIAIts graphics processing unit (GPU) powers most of the large language models developed by OpenAI. MicrosoftGoogle and Amazon And it also supports the heaviest artificial intelligence workloads.
Despite its low profile, Broadcom’s accelerator chips, which the company calls XPU, have become a key part of the artificial intelligence ecosystem.
“The reason it’s so hot is because they’re talking about artificial intelligence, artificial intelligence, artificial intelligence, artificial intelligence,” Eric Ross, chief investment strategist at Casscend, told CNBC’s “Squawk Box” earlier this month.
Intel, which has been the dominant U.S. chipmaker for decades, has been all but shut out of artificial intelligence. Its server chips lag far behind Nvidia, and the company has lost market share to long-time rivals AMD At the same time, huge sums of money were spent to build new factories.
Pat Gelsinger was ousted from Intel’s board of directors on December 1 after a tumultuous four-year tenure.
“I think the more innovative people may have seen artificial intelligence,” Paul Argenti, a management professor at Dartmouth’s Tuck School of Business, told Squawk Box after the announcement. The wave is coming.
An Intel spokesman declined to comment.
Broadcom’s current market capitalization is approximately US$1.1 trillion, making it the eighth US technology company to exceed the US$1 trillion mark. It is the second most valuable chip company behind Nvidia, which is driving the artificial intelligence boom to a $3.4 trillion valuation. apple among all listed companies. Nvidia stock has soared 178% this year, but has actually performed better in 2023, when it rose 239%.
Until four years ago, Intel was the world’s most valuable chip maker, with a market capitalization of nearly $300 billion at the start of 2020. replaced) and has been negotiating to sell off parts of its core business. Intel Now ranks The market value of global semiconductor companies ranks 15th.
“Not for everyone”
After Avago merged with Broadcom in 2015, the merged company’s largest business was chips for TV set-top boxes and broadband routers. Broadcom still makes Wi-Fi chips used in laptops, iPhones and other smartphones.
After acquisition attempt of chip giant fails Qualcomm In 2018, Broadcom turned its attention to software companies. Its spending spree peaked with the announcement in 2022 of the $61 billion acquisition of server virtualization software provider VMware. Software accounted for 41% of Broadcom’s $14 billion in revenue in its most recent quarter, thanks in part to VMware.
What excites Wall Street is that Broadcom is working with cloud providers to build customized chips for artificial intelligence. The company’s XPUs are generally simpler and cheaper to run than Nvidia’s GPUs, and they’re designed to run specific artificial intelligence programs efficiently.
Cloud providers and other large web companies spend billions of dollars each year on Nvidia’s GPUs so they can build their own models and run artificial intelligence workloads for their customers. Broadcom’s success with custom chips is setting up an artificial intelligence spending showdown with Nvidia as the hyperscale cloud company looks to differentiate its products and services from rivals.
Broadcom’s chips aren’t for everyone, as only a handful of companies have the ability to design and build their own custom processors.
“You have to be Google, you have to be Yuanyou must be a Microsoft or Oracle On December 13, the day after Broadcom reported earnings, Piper Sandler analyst Harsh Kumar said on CNBC’s “Squawk on the Street” that “these chips are not for everyone.”
While 2024 is shaping up to be a breakout year for Broadcom—AI revenue grew 220%—December has already set a new all-time high. The stock was up 45% this month as of Monday’s close, 16 percentage points above last month’s best performance.
During the company’s earnings call on December 12, Tan told investors that Broadcom had doubled its XPU shipments to its three hyperscale suppliers. The most prominent of these is Google, which relies on its tensor processing unit (TPU) technology to train the artificial intelligence software Apple released this year. Analysts said the other two customers are TikTok parent company ByteDance and Meta.
Tan said companies could spend $60 billion to $90 billion on XPUs in about two years.
“By 2027, we believe they each plan to deploy 1 million XPU clusters on a single fabric,” Tan said of the three hyperscale customers.
In addition to AI chips, AI server clusters require powerful network components to train state-of-the-art models. Network chips used for artificial intelligence accounted for 76% of Broadcom’s US$4.5 billion in network sales in the fourth quarter.
Broadcom said that overall, about 40% of the US$30.1 billion in semiconductor sales in 2024 will be related to AI, and AI revenue will grow 65% in the first quarter to US$3.8 billion.
“The degree to which hyperscalers will succeed in this move is clearly an area of debate,” Cantor analyst CJ Muse, who recommends buying Broadcom stock, wrote in a Dec. 18 report. For those taking advantage of custom chips , paying attention here will continue to be a meaningful boon.
Intel had a really bad year
Before 2024, the worst year for the Intel market was 1974, when the stock price fell 57%.
The seeds of the company’s latest setback were sown several years ago, when Intel missed out on the mobile chip market from Qualcomm, ARM and Apple.
Leveraging its efficient manufacturing relationship with AMD, rival AMD is beginning to gain market share in the key PC and server CPU markets. British Semiconductor. For years, Intel’s manufacturing processes have lagged a notch, resulting in central processing units (CPUs) that were slower and less energy efficient.
But Intel’s most expensive business is artificial intelligence—a big reason why Gelsinger was fired.
Originally designed for video games, Nvidia’s GPUs have become key hardware for developing power-hungry artificial intelligence models. Intel’s CPU, once the most important and expensive component in a server, has now become a secondary component in AI servers. The GPUs Nvidia will launch in 2025 won’t even require an Intel CPU – many of them will be paired with Nvidia-designed ARM-based chips.
With Nvidia reporting revenue growth of at least 94% over the past six quarters, Intel has been forced into layoff mode. Sales have declined in nine of the past 11 periods. Intel announced in August that it would lay off 15,000 people, accounting for about 15% of its total workforce.
“We are working hard to create a leaner, simpler, more agile Intel,” Chairman Frank Yeary said in a press release announcing Gelsinger’s departure on December 2.
A big problem for Intel is the lack of a comprehensive artificial intelligence strategy. It touted the artificial intelligence capabilities on its laptop chips to investors and unveiled an Nvidia rival called Gaudi 3. Intel’s Gaudi 3 sales this year fell short of its $500 million target.
At the end of next year, Intel will release a new AI chip code-named Falcon Shores. It will not be built on the Gaudi 3 architecture, but will use a GPU.
“Is this going to be great? No, but it’s done,” Intel interim co-CEO Michelle Holthaus said during a Dec. 12 financial conference at Barclays. A good first step for the platform.
Holthaus and interim co-CEO David Zinsner vow to focus on Intel’s products, abandoning the fate of Intel’s costly foundry unit Not sure.
Before leaving, Gelsinger championed a strategy that would have required Intel to gain a foothold in the semiconductor market and make chips to compete with TSMC. In June, at a conference in Taipei, Gelsinger told CNBC that when its factories are up and running, Intel hopes to make “artificial intelligence chips for everyone” and provide an alternative to TSMC for companies like Nvidia and Broadcom .
Intel said in September that it planned to turn its foundry business into an independent unit with its own board of directors and the possibility of raising outside funding. But for now, Intel’s main customer is Intel. The company said it doesn’t expect meaningful sales from external customers until 2027.
At a Barclays event this month, Zinsner said the foundry business’s independent board “has stepped up today.” More broadly, he said the company is looking to eliminate complexity and associated costs wherever possible.
“We will continually review where we spend our money to make sure we are getting an appropriate return,” Zinsner said.
watch: Intel plans to take chip subsidiary Altera public
About The Author
