The Age of Compromises: A Data-Driven Look at the End of Annual Flagships
Smartphone OEMs may soon face real obstacles to making systematic improvements in their lineup year after year.
Most of us here are Android power users and enthusiasts. We take for granted that smartphones will get better and better every year—we demand ever more from smartphone OEMs, and we get upset when we don’t see enough improvement on the latest flagship year-over-year. But the past year in the industry has been disappointing for many. Though there are a couple of exceptions, most 2015 flagships have failed to ignite the passion and excitement as those of yesteryear.
The primary problem, this year, is that there have been compromises—compromises everywhere. Many OEMs are choosing the Snapdragon 808, which is scantly better (if at all, especially on 1440p) than last year’s 801, because the 810 has overheating issues; internal storage capacity remains uncomfortably low on virtually all phones, micro SD slots keep disappearing, and additional capacity is so expensive you might think it’s still the year 1992; thinness is still widely-prioritized over battery capacity, while few OEMs choose to have a removable battery; and every OEM insists on adding “value” to Google’s expertly-designed operating system by dumping bloat and ugly skins on top, while the one OEM that delivered a near-stock experience is currently being gutted by Lenovo.
There seems to be no perfect phone, this year. In fact, I can’t think of any almost perfect phones, either. For those of us still using 2014 phones—or hell, even 2013 flagships—the minor boost in specs doesn’t seem to justify the major expense of upgrading to a shiny new smartphone.
With this backdrop, I’d like to take an opportunity to discuss the C-word—compromise—and why we may be seeing more of it in the future.
We’ve previously flirted with the idea of a coming “age of compromises,” but the fact is that smartphones can’t keep improving at the pace that we’ve been seeing—and expect—indefinitely. Academics and engineers have been openly discussing the end of Moore’s Law in less than a decade, but smartphone OEMs are additionally limited by myriad practical considerations. From physical size and battery life to price and performance, smartphone OEMs may soon face real obstacles to making systematic improvements in their lineup year after year. At a bare minimum, annualized smartphone releases may be a thing of the past. Even Apple—once well-known for its annual product cycle on various products—has slowed down. In the worst case, though, hardware stagnation potentially looms before us.
To better see the long-term trends of the industry, I’ve organized data on all of the major OEM flagships—Samsung, LG, HTC, Sony, Motorola, and Nexus-branded devices—and prepared figures. Let’s take a look.
We start off with benchmarks1. If we take a look at the above figure, benchmarks are in a solid upward trajectory from 2011 through 2014… but then something happens that crashes scores across the board—that something is the Snapdragon 810. Probably due to the widely-publicized overheating issues and the throttling put in place, the Snapdragon 810 performs poorly in benchmarks (where CPUs are taxed to their limits… and heat up). Both Sony and HTC used the SD810 in their latest flagships, and they lie at the bottom of the pack in these CPU benchmarks. LG and Motorola both fare better with the SD808, but neither perform as well as their SD801 predecessors from 2014. Samsung is the outlier, here, as they used their internally-developed Exynos SoC. Though often panned in the enthusiast community for being the odd man out, it looks like Samsung may have made a good choice to stay away from the latest family of Snapdragon processors.
Even if mobile CPU performance continues to improve in benchmarks, it’s an entirely different matter whether or not that’ll translate into noticeable real-world performance.
Unlike many of the other technical specs of smartphones, the CPU is one area that can, presumably, continue improving for quite some time. Maybe Moore’s Law will come to an end within a decade, but in the smartphone industry, a decade is an eternity. Unlike non-mobile platforms, smartphone OEMs have energy consumption and performance per watt to consider, and mobile CPUs are far behind those of desktops and servers in terms of performance. It’s because of this, though, that there remain numerous avenues for improvement on mobile CPUs for years to come. It’s likely that even long after Moore’s Law comes to an end and CPU performance plateaus, mobile devices will continue to see performance gains through improvements in both hardware and software efficiency.
A bigger question, though, is whether further improvements in mobile performance are necessary or useful. Even on the desktop, CPU performance is rarely a bottleneck; operating systems are light on system resources, and few games, even, push CPUs to their limits. Even if mobile CPU performance continues to improve in benchmarks, it’s an entirely different matter whether or not that’ll translate into noticeable real-world performance gains—well, unless people start playing Crysis on their phones.
1 Note that for the purposes of these comparisons, it was important to find a benchmarking methodology that consistent from 2011 until present. All of the most popular benchmarking software, such as AnTuTu and 3DMark, have undergone major revisions in that time, resulting in substantially different scores depending on the version used. AndEBench, while not popular, was developed by an industry consortium to provide a consistent performance benchmark over long periods of time. Scores were retrieved via the AndEBench site and Notebookcheck.com.
RAM is trending upwards, but slowly. Unlike CPUs, the primary reason we don’t see larger numbers, here, is economic. There’s nothing stopping Samsung from throwing more RAM in the Galaxy S6, but it was only a year and a half ago that they chose to regress back to 2 GB even after the Note 3 was released with 3 GB RAM. Most laptops, today, generally sport 4 GB of RAM, and even the Microsoft Surface 3, which runs full Windows 10, starts with 2 GB. The fact is that RAM growth on desktops and laptops has been stagnant for years, and people do less on their mobile devices than they do on dedicated PCs. Nevertheless, most flagships are now announced or rumored to have 4 GB, so the steady upwards creep in RAM continues.
DDR4 is the next step (Snapdragon 810 supports it, but the 808 does not; newer models almost certainly will). While DDR4 certainly has advantages over DDR3 (lower voltage, for example), don’t expect much in the way of performance improvements. RAM is virtually never a performance bottleneck, and comparisons between performance with both DDR3 and DDR4 RAM are really nothing to write home about. Will DDR4 improve benchmarks? Probably. Will it improve real world performance? Doubtful.
Screen size seems to just keep going up and up. It can’t go up forever, though; there is a limit, and I would suggest that we’ve hit it. Though I can’t seem to find any market research on what screen size the public prefers, there are some design decisions by major OEMs that should give a decent barometer of public opinion. For instance, when Motorola was bought by Google and challenged itself to design the ideal Android smartphone for the masses, it chose a 4.7″ screen. When Samsung internalized the complaints about the widely-panned Galaxy S5 and reinvented the series with the Galaxy S6, it chose a 5.1″ screen. And Apple—which, of all companies, you would assume knows something about consumer preferences and marketing—only recently introduced a phone larger than 4.7″–and still, the smaller iPhone 6 continues to greatly outsell the iPhone 6 plus.
Screen PPI is on an upward trend. But once, again, you can only go so far, and we’re pretty close to the end—or at least an extended plateau. “Retina Display” is generally in the range of ~320 PPI, and the highest-density screens currently on Android belong to the Galaxy S6 at 577 PPI. How much higher will PPI go? Just last year, Sharp unveiled a concept screen with 736 PPI, which would meet or exceed the limits of the human eye. If we keep going up, taken to its logical conclusion, an 8K resolution on a 12″ tablet would be about 747 PPI, while a 4K resolution on a 6″ phablet is approximately 734 PPI. However, 2K screens are still not yet ubiquitous on flagships, 4K only exists in the realm of ultra-premium TVs and monitors, and 8K doesn’t even exist, yet, outside of research labs and expos. Little to no content exists at these resolutions to justify both the additional expense and the burden on the CPU and GPU, neither or which are trivial.
Like screen size and screen pixel density, screen resolution is on an upward trend. Resolution can’t increase independently, though. Screen resolution is both a function of screen size and PPI, and it’s impossible to talk about one without talking about the other two.
We’ll be done with pixel density, and thus we’ll be done with increasing display resolutions… unless VR picks up.
Just a few years ago 720p was considered reasonable, and at the screen sizes of the time, a 720p screen was at “Retina Display”-levels of pixel density (~320+ ppi). Since then, screen sizes have gone up, not only due in part to consumer demand, but also due to the ability of manufacturers to create denser and denser screens. Display resolution will rise as pixel density improves and screen sizes get larger. After all, if you can manufacture a screen at some amount of pixels per inch, it’s trivial to just manufacture more inches of it. Screen sizes are typically standardized, though, so increases in screen size will follow predictable jumps: HD (1280×720), Full HD (1920×1080), Quad HD/2K (2560×1440), and Ultra HD/4K (3840×2160). Screen pixel density is not standardized, though, so expect these resolutions to appear with various screen sizes and screen PPI’s.
However, if screen sizes stop increasing (which I think they have) and pixel density stops rising (we’re very near the limits of the human eye), display resolution will cease to increase. If the industry continues at its current rate, we can expect to see UHD/4K screens on flagship phones soon. At that point, pixel density will exceed that which the eye can see, and there will be no more reason to improve. We’ll be done with pixel density, and thus we’ll be done with increasing display resolutions… unless VR picks up.
Here’s a chart that, at first glance, seems to be encouraging. Battery capacity is in an upward trajectory. I’m sure everyone who has ever owned a smartphone has, at some point in time, wished they had a bigger battery. Although sometimes it seems that OEMs choose fashion over battery (and it’s obvious from the figure which OEMs are habitually behind the curve), it’s pretty clear that capacity is consistently improving every year. Battery life isn’t strictly proportional to battery capacity, though. On top of that increased physical capacity, every new iteration of Android brings with it improved efficiency and new battery-saving features; individual OEMs, too, work to optimize their phones for battery savings as best they can.
There is a recurring theme in this article, though. Yes, it’s nice that battery capacity for flagships is now upwards of 3,000 mAh, on average, but how much farther can we continue to go? At some point we’ll hit the ceiling, where added capacity doesn’t justify the extra weight and size. Additionally, battery capacity is almost certainly related to screen size—as screen sizes go up, so, too, do the physical dimensions of the phone; a larger screen means more physical space to put more battery. With screen sizes hitting their sweet spot and without radical changes in battery technology, OEMs have only one way to add more capacity—add thickness. So are they doing it? Well, we have a couple figures for that.
Firstly, let’s take a look at the volume—the physical space inside of a smartphone, inside of which OEMs must fit all the components—of flagships over time. The physical space inside of a phone is certainly increasing over time, but how much of that is due to larger screens and how much is due to added thickness?
I previously mentioned that battery capacities may be piggy-backing off of the ever-increasing screen sizes on mobile devices. But with screen sizes stabilizing at between 5″ and 6″, OEMs are going to have to start adding thickness to smartphones if they wish to continue to increase battery capacity. Is there any evidence of that happening? Take a look at the next figure.
The most notable thing, here, is the absence of any sort of industry-wide patterns. Some flagships are getting thicker over time, while others are getting thinner. Further, there’s not really any correlation between thickness and battery capacity. Companies like Sony routinely pack large batteries in their phones, but they are at the bottom of the thickness charts; conversely, Motorola ranks number one for thickness, but their batteries are small.
So if thicknesses aren’t really increasing—and even when they do, they don’t correlate with added battery capacity—the conclusions are clear: the added physical volume within smartphones that has aided manufacturers in adding ever-larger batteries is primarily driven by the growth in screen sizes. With screen sizes pretty much fixed for the foreseeable future, short of thicker phones (unlikely) or a breakthrough in energy storage density (even more unlikely), we’d better start getting comfortable with the ~3,000 mAh range.
I think this is an interesting figure. How many of you think about mass when phone shopping? The fact is, phones have been getting heavier over time—to the tune of nearly 30% over 4 years! When you think about it, though, it makes sense. Screen sizes have been going up, and so the physical dimensions of phones have been going up, too—that means more material and more weight. Battery capacity is also increasing, which adds a lot more weight. In addition, phones, in general, have been becoming more premium over time—in 2011, aluminum accents were an anomaly; today, they’re typical. All of these things add weight. But since screen sizes probably won’t increase much more, and OEMs haven’t been keen to add more to the thickness of phones, the steady increase in phone mass over time is probably going to be a thing of the past.
Minimum storage is something that should have gone up a long, long time ago. The price per GB of NAND storage has been dropping precipitously for years, but smartphone manufacturers loathe to pass those savings on to consumers. Introductory storage space has remained at 16GB for most phones for five years. We’re still getting ripped off with $50 to $100 price tags for upgrading to larger storage capacities, and microSD slots are something of an endangered species. This is the only spec in the whole industry that has refused to budge; we’re finally starting to see an inkling of a new 32 GB minimum standard, though, and it’s about damn time.
Rear Camera Resolution
Megapixels have been trending upwards, but the scatter is higher than in most other charts. There’s a good reason for that: design—both hardware and software—are more important in determining the quality of a digital photo than the absolute resolution. For instance, the Samsung Galaxy S6 and Note 4 both share identical hardware cameras, but the S6 takes better pictures; and newer Xperia phones take better photos if you abstain from unlocking them. Some OEMs have held megapixel counts stable, while others have even decreased the megapixel count from one generation to the next (i.e., HTC). Despite this, image quality has generally improved over time. How many of you remember the terrible image quality of the Galaxy Nexus camera, for instance?
Looking at the trends broadly, we’ve just about hit the ~20 MP mark; even high quality point and shoot digital cameras don’t really go much higher than that. Anything much higher will probably result in an increase of noise, rather than an increase of image quality. The take-home message here is that race for more megapixels is done. The end of resolution improvements doesn’t signify the end of innovation or image quality, though. As mentioned, lens and software design are much more important than resolution, and we aren’t even in uncharted waters, here—the race for more megapixels came to an end in the world of point & shoot digital cameras years ago, and the result was that manufacturers started competing on features, software, image quality, and smart automatic camera modes. We’ve already seen some smartphone manufacturers resort to gimmicks to differentiate their products (dual rear cameras, anyone?), but there are many ways to innovate and differentiate. One such new feature that comes to mind is optical image stabilization, which has been making its way into most flagship phones.
I hope the figures I have presented have been enlightening, and I’m sure that everyone reading will draw their own conclusions. In the absence of any external information, the clear, upward-trending linear growth across nearly every smartphone specification might leave some people with the idea that this growth will continue forever—or at least unabated in the near-term. I’ve come to very different conclusions, however. I think that this clear, predictable growth can’t continue forever, and indeed it will come to an end sooner than we might expect. Us Android enthusiasts have been spoiled by these years of high spec growth rates, and we’ve even grown to expect such improvements year after year. Some specs, such as phone size and rear camera megapixel count, have reached the end of their growth, while in other areas continuous improvement will become exponentially more difficult over time as OEMs reach the limits of current technical capability.
Although I think it’s likely that we’ll continue to see new, exciting, and innovative phones for a long time to come, I do think that the idea of concrete improvements year after year after year is going to face an abrupt death. In my opinion, the annualized smartphone model is over, or will soon be over. Like with Apple, I think that we’ll soon start seeing major flagships receive only minor upgrades or additions annually, while complete redesigns will take two years or more.
Below are my predictions for the future. Do you agree with them? Let us know in the comments!
- CPUs will improve, but does it matter? Mobile CPUs are limited by heat, efficiency, and price considerations. As Moore’s Law slows down, phones will no longer see the performance, heat, and efficiency gains merely from using smaller processes. In addition, we may already be at a point where the CPU is strong enough that more speed won’t necessarily mean a snappier operating system. If that’s the case, further performance gains won’t result in noticeable real-world improvements, and so new CPUs won’t excite like they once did.
- RAM will increase, but slowly. Smartphones are rapidly approaching parity with laptops in terms of RAM capacity. RAM is expensive, and OEMs’ very slow rate of RAM capacity improvements shows that they’ll continue shipping phones with as little RAM as possible; they won’t include more if the performance boost doesn’t justify it.
- Screen size has stabilized between 5″ and 6″. Any larger and smartphones will have become tablets. Sales of current phone models also show that consumers prefer moderately-sized phones to behemoths like the Nexus 6. There will always be a market for phablets, but the median size of smartphones should grow no more.
- Pixel density is nearing its logical conclusion. The pixel density on flagships is already upwards of 500 ppi, and the Galaxy S6 is approaching 600. Sharp has already developed a prototype screen with 736 ppi, which exceeds the limits of what the human eye can resolve. Once you reach that point, there’s not much reason to go further.
- Display resolution will stay at QHD for years. With screen size and pixel density stagnating, resolution will stagnate, too. Even with Sharp’s super-dense prototype screens, you can only fit 3840×2160 pixels (UHD/4K) on a 6″ screen. Smartphones have been a driving force for improved resolution and pixel density in computer displays, but 4K may be the end-game—at least for anything smaller than a large tablet. And at our current pace of resolution growth, it’ll be years before mainstream phones adopt 4K; I expect we’ll mostly be working with QHD screens for the foreseeable future… even if the first jump is near.
- Battery capacity may have reached its limit. We’re finally up around the 3,000 mAh mark for median battery capacity in flagship phones, but there it will probably stay. Ever-growing screen sizes have given OEMs more physical space to shove bigger batteries into phones, but as screen sizes stagnate, so too do phone sizes, and OEMs have shown little interest in adding extra thickness to phones. Thus, unless we see improvements within the batteries, themselves—improved energy storage density, for instance—we’d probably better get used to 3,000 mAh batteries.
- Battery life will continue to improve. Despite battery capacities stagnating, there is plenty of room for battery life improvement. Every new version of Android sips less battery than the last, and from screens to radios, OEMs improve efficiency where they can, too. With CPU fabrication getting smaller and DDR4 running at a lower voltages, we can expect battery life to improve even as capacity growth is at a standstill.
- The megapixel race is done, but that doesn’t mean cameras will stop improving. Rear camera resolutions and sensor sizes equal or exceed dedicated point & shoot cameras; additional megapixels will just add noise. The digital camera industry reached the megapixel finish line years ago, and they have since focused on improving software and features to differentiate from competitors. Software makes a huge difference in image quality, as can be seen by the difference in quality between the Samsung Galaxy S6 and Note 4, despite identical camera hardware. Optical image stabilization is the Next Big Thing, and we’ll likely see more gimmicks as manufacturers strive to differentiate themselves from competitors.