CPUs are a duopoly. While some alternatives to AMD and Intel technically exist, their market share is so tiny that they do not appear on some statistics tracking pages. It is for this reason that we rarely talk about only Intel or only AMD. We instead compare them to one another, as they are both the only true competition to each other.
So, in today’s update article, as we turn our attention to whether there’s a place in our PC build chart for Intel’s strange new CPU release, named ‘Arrow Lake’—we also consider how this release fits into the history of the battle between two impressive (sometimes clumsy) tech giants.
Intel’s Rise to Power
We have followed Intel’s success over the years, starting from the 2006 Core 2 launch, when the earth was young and so were we. The Core 2 launch was Intel’s return to power after half a decade of AMD dominance. You can read Anand’s review of the launch, and note how strong Intel was with that launch. With prices starting at $1000 (which would be inflation-adjusted to ~$1600 today), you could tell that Intel was fully aware that it had produced a winning CPU that would sell at any price. The CPU market share was roughly 50:50 just prior, but Intel started pulling ahead with that victory.
Ah, but ~$1600 CPUs are not for everyone, and Intel was wise enough to follow up with the Core 2 E8000 series in 2008. With excellent single-threaded performance and affordable prices (note especially the budget-friendly and very popular E8400), this launch was well-reviewed, and helped Intel gain market share in more segments.
Intel’s Dominance
UNLIMITED POWER!
After 5 years of strong performance, Intel went for the jugular. In January 2011, the legendary Sandy Bridge CPU series was launched, and Sandy Bridge set the standard for “overwhelming triumph”.
You can get a sense of how incredible Sandy Bridge was by reading Tom’s 22-page review, or you can settle for a summary: cooler, cheaper, less-power-consuming, and enormously (40%!) more powerful than the previous year’s $1000 stuff! If you were buying a new PC at that time, you were quite lucky. Sandy Bridge set the stage for Intel’s firm hold on its king-of-the-computing-hill status that lasted for a full nine years.
Intel was in total control of the CPU market, with cool, strong, cheap CPUs. Unfortunately, Intel was a victim of its own success, which was exacerbated by AMD’s failure to provide strong competition. Intel was far ahead, and AMD was not catching up any time soon, so Intel started coasting. You can enjoy our coverage of the 2006-2017 period, where we documented the agonizing pain of watching AMD struggle fruitlessly while sliding backwards in performance. AMD’s retreat meant that Intel simply did not need to innovate or improve its performance: why put in effort when your opponent is failing without your input?
Intel’s Stagnation
From 2012 onwards, Intel adopted a strategy of very tiny improvements per generation. Some CPU series would have a 5% improvement in performance; some would have a tiny overclock; some would improve the iGPU. No one was happy with this, but what were you to do? AMD had nothing to offer, so it was either Intel’s rebadges or AMD’s vacuum of options. Intel looked at itself (80% market share), looked at AMD’s position far behind (a pathetic 20% market share), and decided to go to sleep.
In 2017, AMD stopped treading water and started chasing Intel with its Ryzen launch. Ryzen was weak in single-threaded performance, but strong in multi-threaded, so it was a “mixed at best” deal. It showed, however, that AMD was serious about a comeback. In response, Intel’s slumber went on apparently undisturbed, happy to continue coasting.
AMD’s New Hope
In 2019, Intel’s relaxed napping was interrupted by AMD’s launch of Zen 2, which saw a competitive and healthy set of AMD CPUs that could stand toe-to-toe with Intel’s lineup. Then AMD overtook Intel with its Zen 3 one year later. Intel’s refusal to improve had caught up with them, and after a full 15 (!) years of being the top dog, Intel was now in second place. Zen 2 was Intel’s wake up call, and Intel went to work.
Limited power.
For the next two years, Intel’s worker bees buzzed away on Rocket Lake, Intel’s next series. Sadly for Intel, and just a few months prior to Rocket Lake’s launch (end of 2020), AMD preemptively launched its Zen 3 CPUs, and they were excellent. AMD had extended its lead on Intel, adding to the pressure.
And thus Rocket Lake launched in early 2021 to high expectations… and missed its target completely!
Called things like “waste of sand” and “pathetic“, Rocket Lake was the object of many memes (14+++++!) and many jokes, and we enjoyed covering its launch. This was Intel’s first fumble in quite a while, and its market share eroded. At that time, Intel’s market share had dropped from a previously overwhelming 80%, to a more threatened position at 60%. Effectively, AMD had doubled its market share from 20% to 40%.
The Intel Empire Strikes Back
Despite the memes and the jeering, Intel soldiered on. The failure of Rocket Lake in early 2021 was brushed off, and in a surprising move, Intel released Alder Lake in late 2021. Alder Lake was a “performance-at-any-cost” type of launch, with serious power draw and high temperatures, but it delivered when it came to performance, and put Intel back on the radar as a competitor.
A year later, Intel looked at AMD’s excellent Zen 4 launch, and answered with its own perfectly-timed and stellar Raptor Lake series of CPUs. That is right: Intel was back! No more sleeping, no more coasting, no more underestimating the competition; just strong engineering and good CPUs. After the Rocket Lake mislaunch, Alder and Raptor had helped Intel claw back 5% market share from AMD, with the split becoming 65:35 in that time period.
The victory did not last long though, as Intel slipped back into the old habits that let AMD catch up. Raptor Lake was great? Well, let us return to the same strategy we used for 5 years: release a feeble “refresh” that no one wants. The Raptor Lake Refresh was released in 2023 to groans and complaints, as this “refresh” was a 2% improvement in performance over its predecessor. No one was impressed, and Intel was once more the object of memes and ridicule. With real competition now on the scene to make the old habits a much more obvious misstep, Intel had fumbled a second CPU launch.
Today’s Dip in the Arrow Lake
Intel aimed to produce an innovation of this caliber.
In late October, Intel launched its latest series of CPUs, Arrow Lake. Much was riding on this launch, as Intel had once again dropped to 60% market share. Even more concerning, AMD had launched its Zen 5 series of CPUs in August, and they were great. At this point, AMD had scored 4 wins in a row, and were becoming more popular than Intel in some areas. Intel needed a strong win, and Arrow Lake… well, Arrow Lake was a very mixed bag.
Let us take a closer look, but before we continue we will list some sources. My hope is that the editor does not find them and move them, giving the reader the rare chance to check on sources placed neither at the very top or the very bottom of an article:
Intel launched the series with 3 CPUs, as follows:
- Intel Core Ultra 5 245K with 6 performance cores and 8 efficiency cores for $320.
- Intel Core Ultra 7 265K with 8 performance cores and 12 efficiency cores for $390.
- Intel Core Ultra 9 285K with 8 performance cores and 16 efficiency cores for $590.
As you can see from the new naming scheme, Intel looked at the old “Core i9 14900KS” naming scheme and decided it was too short, and thus each “i” has become “Ultra”. Please stay tuned for the day when Intel adds the words “Deluxe”, “Extreme”, “Pro”, and “Ultimate”, because marketing fluff is easier than engineering. But speaking of engineering, how are these CPUs? Let us take a look:
The Good
The biggest three changes from the previous generation are the smaller lithography, the chiplet design, and the NPU.
- The smaller lithography (from 10nm to 3nm) means that the processors are much more power-efficient, drawing far less power (even at full load) when compared to last year’s electricity-bill-destroyers.
- Less power draw means lower operating temperatures, and these CPUs run quite cool. Well done, Intel!
- The NPU, or neural processing unit, is a first-time addition to a desktop CPU, though it has been available for mobile much earlier. It is a section of the CPU dedicated to AI, and Arrow Lake’s performance in AI applications is excellent.
- In addition to the above, the smallest and cheapest of these CPUs still packs 14 cores, which (even in 2024) is a LOT of cores. Arrow Lakes’s multi-threaded performance is stellar.
That is a lot of Good, possibly one of the biggest Good sections we have ever written. If only…
The Meh
…if only we could stop with the Good, then these CPUs would be fantastic. But we continue, for there is more!
- A new socket. Ah, Intel, you will never learn. We are used to it, but will never stop pointing it out: Intel likes to change sockets every 2 years, and we have gone from socket 1700 to socket 1851.
- Better iGPU. The new integrated unit is much stronger than the previous generation, roughly double(!) the performance. Unfortunately, the older Rocket Lake iGPU was very weak, so the new Arrow Lake is still mostly meh.
- The chiplet design seems to be the future, but this is Intel’s first attempt. Arrow Lake’s implementation is better than AMD’s Zen 1, but the (overall) performance and price leave much room for improvement.
- The total core count looks amazing: “Wow, 24 cores!”, but the efficiency cores are the majority. When investigated properly, that amazing 24-core CPU is reduced to 8 bigboi cores, and 16 babycores.
The Bad
This is where we prepare to “oof” and perhaps we may even “yikes”.
- In heavily-threaded and AI applications, Arrow Lake is great, but most applications are not AI and not heavily threaded either. In lightly-threaded apps (i.e. games) Arrow Lake’s performance is subpar.
- When compared to AMD’s options, Zen 5 wins overall.
- When compared to Intel’s own Raptor Lake, Raptor Lake often wins in gaming.
- Despite the fact that performance is a mixed bag (or an outright regression), Arrow Lake is priced higher!
Losing in performance when compared to your competition is bad, but it is normal. Losing in performance when compared to your own older products is very bad. But to lose in performance to your older self and yet have the audacity to ask a higher price? That is just illogical.
Arrow Lake’s Conclusion
Arrow Lake is a mixed bag, but it leans more to the bad side. Perhaps the bag contains some fine sterling silver jewelry, but submerged in mucous. That kind of mixed bag.
If your work is exclusively heavily-threaded or AI, then it is a great choice. But what about the normal, typical user? For us, in general, Arrow Lake CPUs are not a good recommendation at their launch price. If you have no brand loyalty and just want the best, Zen 5 is our recommendation. If you strongly prefer Intel, then Intel’s older 13XXX or 14XXX series are better for most purposes (including gaming). Importantly, Zen 5, 13th-gen Intel, and 14th-gen Intel are all cheaper than Arrow Lake, so you get more performance while saving $$$!
It is extremely rare for new launches from major hardware designers to flop. 90% of all hardware launches are good, and we recommend them. But in those rare 10% instances, we do not shy away from pointing out the bad, including the times when AMD and nVidia have failures. If you have been keeping track, Intel is usually very good, but Arrow Lake needed a bit more time in the oven.
Going Forward
At the current performance and price, we will not be recommending Arrow Lake CPUs. If Intel can execute a little “performance magic” with drivers and microcode updates, or if Intel gives us a price cut, we will definitely revisit Arrow Lake in the future.