Intel’s wins, fails, and WTF moments of 2024

The good news for Intel? 2024’s almost over.

Our collection of the highs and lows of Intel’s 2024 will have you reaching for the brandy. They weren’t good. I mean, aside from some of Intel’s mobile chips, what exactly did it do right? Let’s put it this way: when your ex-CEO prays for your company after he was kicked out, it was a bad year.

As we’ve done for other companies in the past, we’ve collected the best, worst, and head-scratching moments from the past year. Get yourself a hot mug of cider or a cold glass of egg nog, and sit down with as we recap Intel’s 2024. And hold on — it’s going to get bumpy.

Crashing Intel 13th-gen/14th-gen CPUs: FAIL

Intel’s year-long saga of mediocrity just would not go away. From a January advisory by RadTools through more intensive investigations that stretched through the summer, Intel’s latest 13th- and 14th-gen desktop chips were plagued by two questions: what was going on, and, more importantly, would my processor be affected?

Eventually, Intel solved the problem of why desktop systems with both chips inside ran the risk of blue-screening: a combination of faulty microcode and elevated operating voltages that contributed to a bug known as Vmin Shift. But once the bug was identified, it needed to be fixed. And Intel ran through patch after patch to try to and finally nail down the problem, which lasted until September.

Right now, there’s no way of telling if your CPU has been damaged — well, except for a crash, of course. Extended warranties helped, of course, but the reputation of Intel’s Raptor Lake architecture was irrevocably tarnished. Thank goodness Arrow Lake was waiting in the wings, huh?

Intel 14th-gen Core HX notebooks: WIN

At least Intel’s Core HX mobile gaming processors succeeded, right? Well, sort of. The 14th-gen Core HX didn’t really offer that much more than the mobile 13th-gen Core HX did in terms of performance, and we all know that the combination of a Core HX and an Nvidia GeForce can product a howling dervish of a laptop.

As far as how the 14th-gen Core HX fared against the Ryzen 8000 mobile, reviews gave the edge to Intel in terms of single-threaded and multi-threaded performance, but handed AMD the crown in terms of power and battery life. The 14th-gen Core HX performed adequately, just without a whole lot of value. But it also offered AMD a chance to start stealing market share from Intel, which it eventually did. It’s not much of a win for Intel, but you have to take what you can get, right?

Intel’s desktop share decline: FAIL

Historically, Intel has commanded about 80 percent of the PC market. Everything that AMD never quite changes that equation. Except that in desktops, AMD is now approaching 30 percent of the desktop market. The Intel faithful may yawn and roll over, content that Intel would have to give up millions of PCs to AMD before it would lose its majority. But cynics may point at the deluge of bad Intel news and conclude that the market has decided accordingly.

In notebooks, however, Intel has maintained the 80-20 ratio. At stake are millions of PCs set to replace older Windows 10 hardware that will go out of support in October 2025 unless customers buy some extra time.

Windows on Arm: WIN

Intel these days is a big fat, target, and Qualcomm is taking aim. So far, however, it doesn’t seem like buyers are necessarily signing up to buy Windows on Arm PCs instead of Intel, surprisingly. (Mercury Research recently reported that combined Apple/Qualcomm Arm processor sales were 10.3 percent, flat with the quarter before.) That might be due to the unexpected success of Lunar Lake or just some old-fashioned arm-twisting under the table.

Still, a Dell XPS 13 powered by a Qualcomm Snapdragon? Embarrassing. A crappy result? Intel saved face, sort of.

With Mediatek and powerhouse Nvidia rumored to be developing Windows-on-Arm processors, though, 2025 may be the year of the Arm PC.

The NPU hype: FAIL

Intel tried to convince us that its first Core Ultra chip, Meteor Lake, would usher in a wave of AI PCs. And they did — but save for Windows 11’s Windows Studio Effects, they really didn’t do too much.

So what was the NPU good for? Two things, apparently: to highlight that Qualcomm’s Snapdragon X platform had an NPU that it was actually shipping, and to usher in the Core Ultra 200 / Lunar Lake, which finally had the horsepower to qualify as a Microsoft Copilot+ PC. If you bought into the AI hype with Meteor Lake, you probably ended up with a sour taste in your mouth.

At least for now, Copilot+ PCs offer just a handful of generative AI improvements within Windows 11. Recall, the feature expected to get customers slavering for local AI, still hasn’t officially shipped. Meanwhile, while both the Core Ultra 200 and the Ryzen AI 300 technically qualify for Copilot+ status, it’s only on paper. Both require Microsoft to issue Windows updates to enable Copilot+ features — though just one, Recall, has been enabled…and that’s just in preview. It’s still a huge mess.

Intel’s GPU situation: WIN and FAIL

Complicating Intel’s NPU issue was the recognition that while NPUs were more efficient at AI than anything else, there was another PC component that could execute AI far better: the GPU. If you owned a PC with a GPU inside of it — especially a desktop — you already owned the most powerful AI engine in the PC ecosystem. The problem? Intel didn’t have one.

The delays between Intel’s first-generation “Alchemist” and second-generation “Battlemage” parts was so big that Intel essentially fell out of the desktop GPU market. Yes, Battlemage (in the form of the Intel Arc B580) is here, it rocks, and its $249 price point will prove to be a boon for cash-strapped PC gamers. But as of this writing both AMD and Nvidia are expected to release far more powerful next-gen GPUs at CES in January 2025, including Nvidia’s Blackwell/GeForce 5000. All either company has to do is decide to release a stripped-down version that competes with Intel on price, and Intel could be toast — though that is unlikely to happen until later in the year, if the competition is even able to compete with the Arc B580’s spectacular value.

Another plus? Intel’s GPU driver issues are largely a thing of the past now.

Intel Core Ultra 200 / “Lunar Lake”: WIN

Let’s be real: Qualcomm is only competing against Intel. AMD is too. But Intel has to compete with both companies, on two different metrics: performance and low power. Lunar Lake, or the Core Ultra 200 series, is Intel’s response to the power-sipping Snapdragon X Elite. And Lunar Lake did very well, offering killer, competitive battery life with Qualcomm’s best. While the CPU performance was middling, Lunar Lake’s gaming performance certainly improved over the prior generation.

What was impressive about Lunar Lake was that it competed with the Snapdragon while also competing with the Ryzen AI 300 — and without any lingering compatibility issues associated with Windows on Arm. I think Matt Smith’s gaming review echoed my own conclusions: Lunar Lake should really be evaluated on a notebook-by-notebook basis, as the performance has been a little inconsistent. Overall, however, Lunar Lake debuted impressively.

Intel “Arrow Lake” desktop (Core Ultra 9 285K): FAIL

When you hold a post-mortem interview acknowledging that the Arrow Lake launch “didn’t go as planned,” you can’t help but call the chip a failure. “I will say that the performance we saw in reviews — to be very clear, through no fault of reviewers — was not what we expected, not what we intended,” Intel vice president and general manager Robert Hallock told Hot Hardware.

Actual users could see even lower performance, too, as Intel’s testing assumed some tweaked settings. (It wouldn’t be the only time during 2024 that Intel was linked to some benchmark hanky-panky, either.)

Arrow Lake’s performance in gaming matched or declined compared to the 14th-gen Core chips, while not delivering amazing gains in terms of power consumption. Contrast Arrow Lake’s disappointing launch with AMD’s outstanding Ryzen 7 9800X3D chip — which obliterated Intel in terms of performance and power — and it was a miserable 2024 for Intel’s desktop chips.

Thunderbolt 5: FAIL, for now

Part of my job requires me to test Thunderbolt docks for a living, and Thunderbolt 3 and Thunderbolt 4 have been quite successful. But Thunderbolt 5 hardware simply hasn’t been as widespread as I anticipated, and a bad first experience with Thunderbolt 5 left a sour taste in my mouth. There’s been a noted lack of Thunderbolt 5 docks, and laptop makers haven’t exactly rushed to include Intel’s discrete Thunderbolt 5 controller, either. Part of the blame must be left at the feet of Intel, which still hasn’t integrated a Thunderbolt 5 controller into a laptop chipset.

The other Thunderbolt flop of 2024 is probably Thunderbolt Share, a cool concept weighed down by licensing and just general unavailability. I’ve seen a couple of Thunderbolt Share docks, but few if any notebooks have adopted the technology. Intel has some work to do here.

Intel’s massive layoffs: FAIL

Let’s face it: 2024 was a pretty lousy year for Intel. Chip nerds had been circling the company for some time — persistent GPU delays, desktop chip problems, and suspicions about the value of AI swung enthusiast perception away from Intel to AMD earlier in the year. But on August 1, Intel announced it would lay off 15,000 workers or so and (*gasp*) eliminate its corporate dividend. Wall Street painted a big red bullseye on Intel’s back and took aim. Public perception changed overnight.

On Halloween, Intel pulled off a tried-and-true strategy: dump all of the bad stuff into a single quarter’s worth of results, losing $16.6 billion after charges on $13.3 billion in revenue. Intel’s stock price plunged. We didn’t know it at the time, but there would be further consequences, too.

The CHIPS Act: WIN

In 2022, President Biden signed the CHIPS and Science Act into law, providing $280 billion for domestic chipmaking and related activities, of which $52 billion was specifically marked for manufacturing and research. According to this nice breakdown of the CHIPS Act funding by our friends at Computerworld, Intel is scheduled to receive $8.5 billion (now $7.86 billion, plus $3 billion in additional funding for participating in the Secure Enclave program as well.) That agreement was signed on Nov. 26, 2024.

That was great news for Intel and then-chief executive Pat Gelsinger. Free money! Subsidies to build out an American foundry business to rival Taiwan and TSMC! But it came with a catch: to receive it, Intel has to hold on to at least 50.1 percent of its foundry business or give up the funding. No problem, right? Well…

Intel’s manufacturing business: FAIL

Intel chief executive officer Pat Gelsinger took a risky gamble almost as soon as he took office: Regain Intel’s manufacturing leadership by jamming through five manufacturing process node upgrades in four years. If Intel can pull off manufacturing Panther Lake on its 18A process in 2025, it will have succeeded.

But Gelsinger also pledged to open a separate foundry initiative to manufacture chips, essentially becoming a competitor to TSMC. The problem? No one bought the concept. Wall Street grew impatient waiting for Intel to fill its foundry business, though Microsoft signed on in February for an 18A chip and the company has claimed other, undisclosed customers have signed checks. Now, who knows what the future holds.

Should Intel have waited until after its manufacturing turnaround had been established to launch a foundry business? It certainly appears so. Meanwhile, Intel built several tiles within both Core Ultra chips at TSMC, rather than its own fabs, while collecting CHIPS Act checks from the federal government. Not a great look.

CEO Pat Gelsinger steps down: FAIL

As a cub reporter, I interviewed Gelsinger when he was in charge of Intel’s USB efforts. He oversaw the Intel Developer Forum. I’ve had a chance to ask questions off and on over the years, and liked the guy. But he never sat down for an interview with PCWorld, and now he’s gone. Learn your lessons, folks.

Gelsinger stepped down just over a month after a blistering Reuters expose pointed out his missteps, including ticking off TSMC founder Morris Chang, who called Gelsinger a “very discourteous fellow.” Insiders said that Gelsinger had a choice: retire or else.

In my opinion, Gelsinger failed for two reasons: He missed the opportunity to challenge Nvidia in enterprise AI, and the Intel Foundry initiative hasn’t taken off so far. But resuscitating Intel’s process technology roadmap should pay off in a big way. The man bled Intel blue, no question about it.

Now, the torch has been passed to Michelle Johnston Holthaus as co-CEO and head of the newly formed Intel Products business. Holthaus has sat down for an interview with PCWorld, so she’s off to a great start.

Hope your new year is happier than Intel’s 2024.