The Qualcomm Snapdragon 865 (7250) SoC is a two-chip affair without an integrated modem, and Qualcomm isn’t going to allow handset manufacturers to use the SoC unless they sell it in a 5G configuration. This has sparked some criticism of Qualcomm and how it’s pushing 5G forward in 2020. Qualcomm has had a rough 2019 — the company was found to have abused its market power in a federal court decision earlier this year (Qualcomm has appealed that ruling).
At Ars Technica, Ron Amadeo argues that the Qualcomm Snapdragon 865 is a demonstrable step backward for smartphone design for a number of reasons, including the aforementioned lack of an integrated modem. He rightly points out that existing 5G hardware overheats if ambient temperatures exceed 85F, the first wave of handsets offers miserable performance, and that 5G as a service is barely available in the United States at all. It’s also true that the few devices to offer 5G service this year were significantly larger than the LTE hardware, and that the last time Qualcomm tried to snap-push a major standard (64-bit ARMv8 support), what we got was the Snapdragon 810, an SoC pretty much nobody liked.
These are all fair and accurate points and I’ve argued in favor of all of them here at ET, save for the idea of a connection between Snapdragon 865 and the ill-fated Snapdragon 810. But I’m not sure all of these trends can be laid solely at Qualcomm’s door. AT&T, T-Mobile, and Verizon have all been spending a great deal of money on 5G deployments (and, in AT&T’s case, blatantly false advertising). The carriers are pushing for 5G service to rollout on devices. OEMs, similarly, want 5G phones, including Apple. Apple felt it was important enough to have 5G in its devices that it didn’t just buy Intel’s modem business — it explicitly settled a huge set of lawsuits with Qualcomm to make certain it would be able to ship a 5G device next year.
5G, in other words, isn’t just a technology that Qualcomm wants to ship. It’s a technological capability that Qualcomm’s partner companies want to be selling and promoting. Qualcomm is responsible for choosing not to sell the 865 with an integrated LTE modem, but it may have made that choice because OEMs weren’t interested in splitting their high-end products into LTE versus 5G buckets any longer than absolutely necessary. Amadeo argues that this integration could drive higher prices and he might be right about that — but if OEMs are looking to push the envelope on high-end handsets, even this may be a feature rather than a bug.
None of this makes 5G a good deal right now for you, the consumer. T-Mobile customers who can use its sub-6GHz 5G service can pick up a modest performance bump today, with more service rolling out in 2020. Those of you who live in dense urban environments might pick up some 5G mmWave benefits as soon as 2020, at least while downtown. The realistic window for 5G to start mattering for most people is probably 2022 – 2023. Depending on where you live, you may not notice much when you get it. But Qualcomm isn’t the only company driving 5G into the marketplace — not by a long shot.
What About the Snapdragon 865, Specifically?
The fact that the Snapdragon 865 is a dual-chip part does raise concerns about the overall impact on power consumption, but Qualcomm is specifically saying they’ve changed aspects of the 865 design to address this. The reasons for not integrating a 5G modem into this product cycle relate to the difficulty of building a full-featured modem to support every single type of 5G network currently being deployed and the lead times associated with bringing up new modems. While Qualcomm did integrate a 5G modem into the lower-end Snapdragon 765, the X52, it’s not as fast as the X55 modem Qualcomm is including with the Snapdragon 865. The X52 has a maximum download speed of 3.7Gbps with 1.6Gbit upload, while the X55 can maintain up to 7.5Gbps download with a 3Gbps peak upload.
We don’t know yet what the practical differences will be, but it’s possible that the X55 will turn out to be a better overall device for the early 5G networks the United States has to offer in 2020. That doesn’t make 5G a good deal as a near-term investment, but it may make the X55 modem the best bad deal if 5G service is what you want.
Ultimately, I’m not as convinced as Ron that the Snapdragon 865 will be a bad part. It’s absolutely true that 5G has made a bad first impression for a host of reasons, but the original LTE phones were downright terrible, and LTE today is a must-have feature. I’m dubious that mmWave 5G will hit “must-have” status in the next 3-5 years, but 5G deployments below 6GHz spectrum could still deliver modest improvements for customers.
If integrating 5G leads to significantly more expensive and larger devices that consumers are unwilling to pay for, Samsung, Apple, MediaTek, Qualcomm, and the carriers themselves will have to collectively deal with that problem. If power consumption and heat are issues, additional node shrinks and further improvements to packaging, antenna placement, SoC interconnects, and device panel power consumption might be needed to bring handset battery life back up to where consumers expect it to be. It’s entirely possible — I’d go so far as to say it’s likely — that it’ll take a generation or two for 5G to really, cleanly differentiate itself. But that’s an issue as much on the carriers as it is on Qualcomm.
I think it’s better to wait and see on the Snapdragon 865 rather than preemptively declaring that it’s going to be a phone wrecked, and that the marketing bonanza around 5G is being driven by a lot of companies. Qualcomm isn’t the only company pushing a 5G-heavy narrative, and while it’s been found guilty of abusing monopoly power, it still has to answer the demands of its own customers, who very much want to be deploying 5G. There’s no shortage of blame to go around as far as sources for bad 5G arguments.
But since I’ve no particular qualm about predictions, I’ll make one of my own. I suspect the Snapdragon 865 will be an improvement over the 5G devices we got last year (no difficulty) but probably will have some issues related to the cost of 5G support.
Battery life may still be slightly lower. The impact of actually using 5G may be high, and it’s possible some devices won’t be able to maintain 5G connections under sustained heavy load and high external temperature. I think Samsung, Apple, and other OEMs may have to find power savings elsewhere in their devices to compensate for the modem, but I suspect they’ll find ways to offset at least some of the increase. OEMs want to sell you on 5G even if you don’t need it, but they’re conscious of the fact that people won’t want the feature if it makes using the device significantly worse in other respects. The goal will be to minimize the impact of 5G support when you aren’t using it, the better to wow you with it when the feature is available.
It’s possible that the Snapdragon 865 may fail at this, and that the devices that use it will be markedly worse than previous-generation LTE products in multiple respects, but I want to wait and see hardware before I make that call. The potential negatives of using an off-die modem can be balanced by improvements in other metrics, and we should see improvements compared with previous hardware. If we don’t, you can bet I’ll have something to say about it.