- Intel is in a predicament, but the company still accounts for huge chunks of the PC and data center CPU markets
- Analysts said it has been too slow to move on AI and needs focus
- Could AMD fall into a similar pit?
Intel is sailing through some stormy seas at the moment. The company posted a $1.6 billion loss for Q2 2024 and announced plans to lay off 15,000 employees. But is the company really a sinking ship, or will its sheer size and incumbent position in key markets keep it afloat despite serious missteps?
Ryan Shrout, president for Futurum's Signal65 division, told Fierce the question of whether Intel is “too big to fail” or not is an interesting one.
“Even though they appear to be so far behind in the world of technology based on their earnings report and the race versus Nvidia in the AI space, you have to keep in mind something like 80% of the client market – laptops and PCs – use Intel chips,” he said. “Even in the data center CPU space 70% or so are using Intel Xeon processors. If Intel disappeared tomorrow, nobody has the capacity to fill that gap.”
AI amateur
While Intel’s incumbent position could protect it for a little while, there’s no question that artificial intelligence (AI) will be infused into nearly all the technology of the future. Thus, Intel’s laggard status on this front poses a big problem.
It’s true, the company has made a play in the AI realm with its Gaudi chipset, but it has yet to make a dent in the market. Shrout said this is partly because Gaudi isn’t a traditional general purpose GPU product that can serve as a simple drop in for Nvidia’s chips. And Intel hasn’t yet woken up to the fact that the market is no longer doing the legwork for it on the software front.
Related: Editor’s Corner: Is Nvidia a monopolist?
“Everybody else built their software for Nvidia because they know they’re the market leader,” Shrout said. “And so that leaves extra work to be done by AMD and Intel if they want to compete.”
Basically, Intel needs to do the work – both on the software front and on the PR front – to make it clear to the market that using the Gaudi chip won’t slow customers down.
How did Intel get here?
While competition certainly played a role in Intel winding up where it has, AvidThink founder Roy Chua told Fierce Intel’s pain is mostly self-inflicted.
“Their lateness in reaching milestone process nodes in past years, over-focusing on business and financial outcomes at the expense of technology investment, missteps on the mobile front, [and its] inability to build a significant software play,” all played a role in forcing the company’s recent layoffs and restructuring.
Shrout, who actually worked for Intel for several years before leaving in 2023, agreed. He argued company leaders for too long either ignored or downplayed what competitors like Nvidia were building. By the time they finally saw the light and decided to change course, it was already too late.
“The competition that’s come into the market was allowed to come in because Intel didn’t see the writing on the wall for the AI movement,” Shrout said. “That’s a self-inflicted blindspot.”
Patrick Moorhead, founder of Moor Insights and Strategy, argued the company simply placed a bet on the wrong architecture when it went with ASIC.
“The Gaudi architecture is a more efficient, ASIC (non-GPU) architecture but is less flexible than a GPU,” he said. Moorhead noted rival AMD went with a more flexible architecture, which allowed it to go after the massive data center market. Meanwhile, the big three hyperscalers developed their own ASIC chips, including Trainium and Inferentia from AWS, Microsoft’s Maia and Google Cloud’s Trillium chips.
“Intel nailed efficiency but the hyperscalers built their own,” he said.
NeXtCurve’s Leonard Lee added that Intel doesn’t have the same level of “programmable and integrated accelerator stack” that Nvidia is peddling. See also: our earlier point about getting its act together on the software front.
“If anything, interconnect and networking have put both AMD and Intel in Nvidia's rear view mirror,” he said.
Additionally, Shrout and several other analysts Fierce spoke to pointed to the fact that Intel has been somewhat distracted from technology advancements by its focus on its nascent foundry business.
Turning it around
To get back on track, Chua said Intel needs to go back to basics and focus on their core businesses. He added it also needs to “demonstrate to their customers and the market that they can continue to innovate in their core, while showing what the Gaudi family can do (like Gaudi 3) in competing with some of Nvidia's business and AMD.”
Jack Gold, Principal Analyst at J.Gold Associates, told Fierce it will likely take Intel two or three years to fully recover from their current position.
“They have a lot of new products in the pipeline, but the biggest issue is they don’t have a good AI story right now,” Gold said. “Their GPU efforts that are geared to large model training are well behind where AMD and Nvidia are right now.”
Gold continued: “Their new Arc GPUs are much more capable, but still behind the others. And it will take them another generation or two to fully combine their GPU and Gaudi capabilities into a competitive AI training chip.”
While Nvidia and AMD are its primary rivals, Gold also noted that there’s an emerging cohort of chip companies that includes Cerebras and Groq as well as chips from cloud players like AWS, Google and Microsoft, that Intel will have to compete with on the AI front. Plus there’s a looming threat from Qualcomm and ARM in the PC space that will likely grow over the next two to three years.
Lee agreed Intel basically has two years to right the ship. “Intel's key to restoring leadership across its client and data center businesses will be nailing the delivery of competitive, quality (low defects and bugs) if not leadership products in the next couple of years,” he said. “If products running up to Panther Lake and Clearwater Forest don't sustain unit volumes through share loss or otherwise to support Intel Foundry's 18A when the product groups bring things back home, the wheels of Intel's overall IDM 2.0 strategy start to fall off.”
Could AMD be next?
Asked whether Intel’s peers might similarly be teetering on the brink, Lee said no.
“Intel is in a different situation than its peers largely due to its IDM model,” he explained. “I would not say AMD is in the same boat as Intel given they are not tethered to fabs and burdened with the duty and the massive CapEx to restore U.S. chip manufacturing leadership.”
Shrout concurred. “For AMD specifically, they were later than they should have been but they’ve been moving in the right path longer than Intel and they’ve been more committed than Intel. So I’m not worried about AMD’s longevity or ability to survive and to take market share.”
That said, Lee conceded that there is the risk that others could fall into a pit if the generative AI hype bubble pops. He concluded “the bigger question is who will be positioned how in the aftermath of that?”