Intel’s worst year, Broadcom’s record grows

It’s been a big year for Silicon Valley silicon, but it’s been a rough year for the companies most responsible for the region’s moniker.

intel, The 56-year-old semiconductor maker, co-founded by industry pioneers Gordon Moore and Robert Noyce and legendary investor Arthur Locke, has suffered its worst year since going public in 1971, and its value has plummeted. Lost 61%.

A completely opposite story unfolded. broadcoma chip conglomerate run by CEO Hock Tan, is headquartered in Palo Alto, California, about 25 miles from Intel’s Santa Clara campus.

Broadcom stock was up 111% in 2024 as of Monday’s close, its best performance ever. The current company was created through a 2015 acquisition by Avago, which went public in 2009.

The driving force behind the divergent stories was artificial intelligence. Broadcom jumped on the AI ​​train, but Intel almost missed it. The two chipmakers’ changing fortunes highlight the ephemeral nature of leadership in the tech industry and how a few key decisions can result in hundreds of billions or even trillions of dollars in changes in market capitalization. I am.

Broadcom is developing custom chips google Giant cloud companies such as It also creates the essential networking gear that large server clusters need to tie together thousands of AI chips. Within AI, Broadcom has been primarily influenced by: NvidiaIts graphics processing units (GPUs) power most of the large language models developed at OpenAI. microsoft,Google, Amazon It also enables even the most demanding AI workloads.

Broadcom’s accelerator chips, which the company calls XPUs, are a lesser-known but important part of the AI ​​ecosystem.

“The reason why it’s really going up is because we’re talking about AI, AI, AI, AI,” Eric Roth, chief investment strategist at Cassendo, told CNBC’s “Squawk Box” earlier this month. .

Intel has been a dominant U.S. chip maker for decades, but it has been largely shut out of AI. The company’s server chips lag far behind Nvidia’s, and the company is also losing market share to its longtime rival. advanced micro device While spending a lot of money on new factories.

Intel’s board of directors on December 1st fired Pat Gelsinger as CEO after a tumultuous four-year tenure.

In an interview with Squawk Box after the announcement, Paul Argenti, a management professor at Dartmouth’s Tuck School of Business, said, “A more innovative person might have foreseen the coming wave of AI.” No,” he said.

An Intel spokesperson declined to comment.

Broadcom is now worth about $1.1 trillion, making it the eighth U.S. tech company to cross the $1 trillion mark. The company is the second most valuable chip company after Nvidia, which led the AI ​​boom and reached a valuation of $3.4 trillion, and only a short distance behind it. apple Among all listed companies. Nvidia stock has soared 178% this year, but it actually did even better in 2023, rising 239%.

Until four years ago, Intel was the world’s most valuable chipmaker, with a market capitalization approaching $300 billion in early 2020. The company is now worth about $85 billion and just fell off the Dow Jones Industrial Average to be replaced by Nvidia, which continues to enter the market. We are in talks to sell core parts of our business. Intel currently ranks 15th in terms of market capitalization among semiconductor companies in the world.

After Avago and Broadcom merged in 2015, the combined company’s largest business was chips for TV set-top boxes and broadband routers. Broadcom still makes Wi-Fi chips used in laptops and smartphones like the iPhone.

After failed bid to acquire a major mobile chip company Qualcomm In 2018, Broadcom turned its attention to software companies. The high point of the company’s spending was the announced acquisition of server virtualization software vendor VMware in 2022 for $61 billion. Thanks in part to VMware, software accounted for 41% of Broadcom’s $14 billion in revenue in the most recent quarter.

What’s interesting on Wall Street is Broadcom’s role in working with cloud providers to build custom chips for AI. The company’s XPUs are generally simpler and cheaper to operate than Nvidia’s GPUs, and are designed to run specific AI programs more efficiently.

Cloud vendors and other large internet companies spend billions of dollars a year on Nvidia’s GPUs so they can build their own models and run AI workloads for their customers. Broadcom’s success with custom chips has set off an AI spending showdown with Nvidia as hyperscale cloud companies seek to differentiate their products and services from competitors.

Broadcom’s chips aren’t for everyone, as only a handful of companies can afford to design and build their own custom processors.

“It has to be Google, it has to be Google.” metaMicrosoft or oracle “These chips are not for everyone,” Piper Sandler analyst Harsh Kumar told CNBC’s “Squawk on the Street” on Dec. 13, the day after Broadcom’s earnings.

2024 was a breakout year for Broadcom, with AI revenue up 220%, reaching record territory in December. As of Monday’s close, the stock is up 45% this month, 16 points above its all-time high.

During the company’s earnings call on Dec. 12, Tan told investors that Broadcom has doubled its XPU shipments to three of its hyperscale providers. The best known of these is Google, which relies on Apple’s Tensor Processing Unit (TPU) technology released this year used to train its AI software. The remaining two customers are TikTok’s parent company ByteDance and Meta, analysts said.

Tan said companies could spend $60 billion to $90 billion on XPUs within about two years.

“In 2027, we believe each company plans to deploy 1 million XPU clusters in a single fabric,” Tan said of the three hyperscale customers.

In addition to AI chips, an AI server cluster requires powerful networking parts to train cutting-edge models. AI networking chips accounted for 76% of Broadcom’s $4.5 billion in networking sales in the fourth quarter.

Broadcom said that a total of about 40% of its $30.1 billion in semiconductor sales in 2024 will be related to AI, and that AI sales in the first quarter will increase 65% to $3.8 billion.

“How successful the hyperscalers are in their efforts here is clearly debatable,” Kantar analyst CJ Muse, who recommends buying Broadcom stock, said in a Dec. 18 note. Yes,” he wrote in a Dec. 18 report. This focus will continue to be a meaningful benefit to those using custom silicon. ”

Before 2024, Intel’s worst year on the market was 1974, when its stock price fell 57%.

The seeds of the company’s latest setbacks were sown years ago, when Intel lost out on mobile chips to Qualcomm, ARM and Apple.

Rival AMD began gaining market share in the important PC and server CPU markets thanks to a productive manufacturing relationship with AMD. Taiwan semiconductor manufacturing company. Intel’s manufacturing process has been set back for years, resulting in slower and less power efficient central processing units (CPUs).

But Intel’s most costly mistake was in AI, which is a big reason why Gelsinger was fired.

Nvidia’s GPUs were originally developed for video games, but are now critical hardware for developing power-hungry AI models. Intel’s CPUs, once the most important and expensive component in servers, have taken a backseat to AI servers. The GPUs that Nvidia plans to ship in 2025 don’t even require Intel CPUs. Many are paired with ARM-based chips designed by Nvidia.

Nvidia has reported revenue growth of at least 94% over the past six quarters, pushing Intel into retrenchment mode. Sales have decreased in 9 of the past 11 years. Intel announced in August that it would cut 15,000 jobs, or about 15% of its workforce.

In a Dec. 2 press release announcing Gelsinger’s departure, board chairman Frank Yeary said, “We are committed to building a leaner, simpler, and more agile Intel.” Ta.

A big problem for Intel is the lack of a comprehensive AI strategy. The company touted its laptop chip’s AI capabilities to investors and released Gaudi 3, a rival to Nvidia. However, neither the company’s AI PC concept nor its Gaudi chip have attracted much attention in the market. Intel’s Gaudi 3 sales missed the company’s own $500 million goal for this year.

Later next year, Intel plans to release a new AI chip codenamed Falcon Shores. It will not be built on the Gaudi 3 architecture, instead it will be a GPU.

“Will it be great? No, but the It’s a good first step.”

The fate of Intel’s expensive foundry division remains uncertain, as Holthaus and fellow interim co-CEO David Zinsner have vowed to focus on Intel products.

Before leaving, Mr. Gelsinger supported Intel’s strategy to gain a foothold in the semiconductor market and make chips to compete with TSMC. Gelsinger told CNBC at a conference in Taipei in June that once its factory is up and running, Intel hopes to develop “AI chips for everyone” and offer companies like Nvidia and Broadcom an alternative to TSMC. Ta.

Intel announced in September that it plans to transform its foundry business into an independent division with its own board of directors and the possibility of external financing. But for now, Intel’s main customer is Intel. The company said it does not expect significant sales from external customers until 2027.

Zinsner said at a Barclays event this month that a separate board of directors for the foundry business was “standing up today.” More broadly, he said the company is trying to eliminate complexity and associated costs wherever possible.

“We’re always going to take a hard look at where we’re spending our money and make sure we’re getting the right returns,” Zinsner said.

Attention: Intel plans to go public with chip subsidiary Altera