Has Qualcomm finally secured unscrupulous control over Android flagships?

If early benchmarks are to be believed, Qualcomm’s new Snapdragon 8 Elite is a serious flagship that easily outperforms current Android and even Apple smartphones. Obviously this will worry other silicon developers. For example, it’s rumored that Samsung is rushing to secure more chips for the Galaxy S25, perhaps because it wants to guarantee the absolute best performance for all its next-generation flagships.

We’re still waiting to see if the 8 Elite will deliver the performance it promises in upcoming consumer phones, but even if it drops by a few percentage points, the next-generation Snapdragon-powered phone will still outperform most competitors. It would be much better. This could lead to major changes in the industry.

For the past few years, the Android chipset market has been largely a case of Snapdragon and “other.” It’s not like rival companies like MediaTek or Samsung are making significantly inferior chips. Performance and feature comparisons are often relatively close, and Qualcomm is not always the first in the industry with new features. But for most of the past decade, partnerships, sales volumes, and to some extent public perception have also tilted in Qualcomm’s favor. It’s hard to name many modern flagship smartphones that aren’t powered by Snapdragon.

Reports that Samsung is once again looking for enough chips to power the Galaxy S series with just the Snapdragon 8 Elite highlight two important issues. First, there are rumors that Samsung Foundry’s manufacturing yields are once again in trouble, resulting in too few suitable chips from its cutting-edge manufacturing process. We first saw signs of trouble with the Snapdragon 8 Gen 1, which is manufactured on Samsung Foundry’s 4nm node. The chip suffered from high temperature and performance throttling issues, which appear to have been resolved by moving to TSMC’s rival node in the Snapdragon 8 Plus Gen 1. Samsung’s own 4nm Exynos 2200 was even more controversial.

The following year, Samsung’s Galaxy S23 series became Snapdragon exclusive for the first time. This was disguised as a strategic partnership, but the return to a mixed chipset strategy with the Galaxy S24 and S24 Plus means Samsung’s Exynos 2300 isn’t competitive enough or Samsung Foundry is still running into yield issues. It confirms that he is suffering.

Cutting-edge silicon manufacturing is only becoming more difficult and expensive. The Exynos 2500 is supposed to be a showcase for Samsung’s latest manufacturing processes. If it fails to do so with the S25 series, rival TSMC will have the only commercial 3nm mobile chip on the market, and a manufacturing monopoly is not good news in terms of capacity or price. It has already been reported that the Snapdragon 8 Elite will be much more expensive than the previous model.

We do not know if the manufacturing industry is responsible for the reported disappearance of Exynos. It is also unclear whether this chip is obsolete or not. For example, it could be included in another Galaxy FE handset or tablet. Still, this suggests that even if Samsung could make chips, it would no longer necessarily come close to being competitive with Snapdragon in key performance metrics, which is the second problem.

Sure, classic performance metrics aren’t everything these days, but Samsung’s dual-chip approach makes it highly vulnerable to performance differences between Arm Cortex CPU components and Qualcomm’s custom CPU developments. Naturally, other players will also be affected. MediaTek’s Dimensity has been pushing sophisticated big-core CPU setups to establish itself as a performance leader, with recent wins in China including the vivo X200 and X200 Pro. However, the Ultra version still opts for Snapdragon. Still, the 8 Elite’s massive 40% improvement is almost certain to catch the eye of all manufacturers looking to sell top performance in late 2024 and into 2025.

This puts rivals like MediaTek and Samsung in an awkward position. Will they continue to rely on Arm’s Cortex program for relatively painless CPU development and hope that the next generation or two will catch up, or will they develop more to close the gap with Qualcomm? Will they also pursue the risky path of custom CPUs? All major companies have probably been evaluating the potential of custom cores for a long time, but development is expensive, time-consuming, and there is no guarantee of success. As it turns out, Samsung’s custom Mongoose core produced questionable results. Qualcomm paid Nuvia $1.4 billion in 2021 for custom CPU development building blocks that currently power the Snapdragon X and 8 Elite, giving it a huge head start.

Even though more custom CPUs will hit the market in the coming years, the graphics and custom AI silicon mismatches are still not resolved. Of course, Qualcomm could lose out on price, but its rivals haven’t had significant success with that strategy so far, and there’s no marketing win to be gained by settling for second place anyway.

Perhaps as problematic as Samsung’s Exynos program is Google’s Pixel. As we well know, Google’s Tensor processors are already behind competitors in every metric except AI. Benchmarks suggest the 8 Elite could deliver twice the CPU and GPU performance of Google’s current flagship Pixel 9 series, making it already look outdated, especially for mobile gamers . Tensor G5 isn’t going to be a particularly ambitious chip either, but it will be a big step up over Tensor G4. Frankly, Google’s progress is slow and it’s just trying to shake off the huge amount of design support from Samsung’s semiconductor division.

The overall picture is already taking shape. Buy Snapdragon if you want maximum performance, and stick with Tensor if you want first access to Google’s AI tools. But how long can brands rely on small pieces of bespoke software to bridge the apparently ever-widening gap between their flagship chips and the rest of the Android ecosystem? Don’t bet on it too long. Sho. Enthusiasts, the typical Pixel consumer, will be keen to see the hardware on the other side of the fence.

Google protects itself from what I call the “Snapdragon effect” with more isolated internal development, but relies on AI tools as a consistent differentiator. But it’s not at all clear that consumers are that interested in this technology buzzword. Internally, Google knows exactly what the performance discrepancy is and may have a contingency plan in place. According to documents seen by Android permissionsGoogle is working on a custom CPU core called Orion that could debut in Tensor G6 or later chipsets. Whether this will be very competitive with what Qualcomm and others are producing by 2026/2027 is anyone’s guess. Perhaps Google will eventually revert to Snapdragon if the pressure becomes too much.

That said, I’m not going to oversell the Snapdragon 8 Elite. We’re not yet into consumer phones, so price and power efficiency will be just as important as the chip’s performance potential. Get any of these significantly wrong and we’ll be having an entirely different discussion about Qualcomm’s CPU choices in a few weeks.

That said, the current broad market for Android mobile processors is delicately balanced by two key factors that are undoubtedly under threat. These are the relatively small performance differences between the chips and the selection of competitive manufacturing partners. Unless the industry as a whole takes a step back and realizes that maybe peak performance isn’t that important for modern smartphones, Snapdragon’s already firm grip on its core markets will ultimately make it difficult for the remaining few. There seems to be a growing possibility that this market will disappear. Alternative. Or maybe Arm will kill Qualcomm’s custom CPU project before missing out on the prize money.