Blog Post

Making a T-Mobile iPhone is harder than it sounds

Stay on Top of Enterprise Technology Trends

Get updates impacting your industry from our GigaOm Research Community
Join the Community!

Last week, Apple-loving(s aapl) T-Mobile customers were all in a titter as the carrier’s CTO Neville Ray hinted at CES that the next iPhone might work on the T-Mobile’s network. Ray only told Cnet (s cbs) that the next radio chipset Apple procures for the smartphone will support the AWS bands that T-Mobile uses for its HSPA+ network, but the resulting slew of speculation about a T-Mobile iPhone forced T-Mobile to put out a statement saying it had no knowledge of Apple’s future iPhone roadmap.

Adding a new band to the iPhone may sound like a simple matter, but it’s a much more difficult task than it appears. In an interview with GigaOM, Barry Matsumori, chief marketing and strategy officer for adaptive antenna maker Ethertronics, explained why.

The problem of handset design

Vendors like Apple simply can’t slap a new antenna into a handset and start shipping. There’s about a $1 cost per device when adding a new band, which isn’t small in the low-margin world of handsets, Matsumori said — though it’s a cost Apple can more easily absorb than most. The real cost is engineering. Adding a new band requires a manufacturer to fundamentally redesign the device, building it around its new antenna configuration. “The big issue is the ripple effect it causes throughout the phone,” Matsumori said.

A new band requires not only a new antenna, but also new power amplifiers, filters and a new radio chipset – all of which have to be crammed into the ever-shrinking confines of the handset. The vast majority of space in today’s slim smartphones is dominated by the battery and display. All the other components are stuffed into the niches alongside the battery. But even if a handset designer can find the space for a new band’s components, he faces an even bigger problem: interference.

Any time you add a new emitting component like an antenna, it changes the radio frequency profile of the device, Matsumori said. Each new radio wave-emitting element can interfere with every other element in the device. A designer not only has to find room for the new antenna and supporting elements, but it has to place them in relation to other components so they don’t degrade the overall performance of the device. Ultimately, the more bands a vendor adds, the poorer signals each of those bands will transmit, Matsumori said.

The iPhone already has signal strength problems, as many iPhone 4 users discovered in 2010. That could be a reflection of the number of bands and technologies it already supports, or it could be a decision by Apple to sacrifice radio performance for industrial design. But the bottom line is Apple needs to think carefully if it wants to extend the iPhone to T-Mobile as any tinkering with the design will not only add cost, but potentially degrade performance for all its customers – not just T-Mobile’s. The next generation of smartphones will start using adaptive antenna technologies developed by companies like Ethertronics and SkyCross, which will help solve some of those interference problems. If Apple has the will, it can definitely find a way to technically fit T-Mobile into its schemes. But the even bigger obstacle T-Mobile faces to getting its sought-after iPhone would be a business one.

“It boils down to a simple business decision,” Matsumori said. “Will these guys give me enough volume to justify the expense? Or will [supporting the new band] be a strategic enough decision for me to buy in?”

When will it be T-Mobile’s turn?

Apple justified that engineering expense when it created a CDMA version of the iPhone 4 for Verizon Wireless (s vz)(s vod). Verizon, however, is not only a much bigger operator than T-Mobile; the CDMA iPhone variant had much more exposure beyond Big Red. Sprint(s S), other smaller U.S. operators and international carriers have since picked up the device. Subsequently, the new iPhone 4S supports both CDMA and GSM/HSPA, but in that case, Apple didn’t have to add any new bands to the device, just a new multi-mode radio chip. An iPhone to support T-Mobile’s HSPA+ network would have a much smaller reach — essentially T-Mobile and a few Canadian operators.

Wireless Intelligence projects 38 distinct LTE bands in 2015

When the iPhone 5 comes along, Apple could finally discover that the math adds up. T-Mobile’s Ray is right that the Apple’s next smartphone will almost certainly have AWS antennas, since AWS is where both Verizon and AT&T(s T) will launch a good hunk of their LTE networks. But that doesn’t mean Apple will support HSPA+ on that band. The antennas will be in place, but Apple will still have to procure radio chipsets that recognize T-Mobile’s network on AWS. Considering the huge number of bands and technologies the next iPhone will need — covering new LTE networks as well as legacy 4G, 3G and 2G technologies — there’s a chance T-Mobile’s specialized configuration still won’t make the cut.

In fact, Apple will find it very difficult to continue its strategy of building a single iPhone that works across global networks. My recent GigaOM Pro analysis (subscription required) explores how the fragmentation of LTE into dozens of bands across the world and within the U.S. will create a big dilemma for handset vendors who will be forced to pick and choose which LTE bands they can feasibly support in a single device. The biggest impact of those choices will be felt by the smaller carriers who don’t have the clout of Verizon to get devices made for their LTE networks.

iPhone/drill image courtesy of Flickr user floorvan

40 Responses to “Making a T-Mobile iPhone is harder than it sounds”

  1. What I don’t get then is how people can use unlocked iPhones on T-Mobile’s network. If they can do it, seems like it’d be easy enough to produce one.

  2. Hi! So..there’s still no news if iPhone’s going to accommodate Tmobile?

    My friend’s phone broke and he went/called to Tmo about a replacement. Someone there told him that in a couple of months they (Tmo) will be coming out with an iPhone.

    Great article, and comments, by the way! Makes my head spin a bit but it did make me understand better.

  3. Raynman_ucf

    Nice article, it helps explain “behind the scenes” reasons why some carriers don’t get some phone models.

    I am curious though, T-Mobile’s UK site has the I-phone for sale. What is different between the T-Mobile UK and US networks? I presume that a UK T-Mobile customer’s phone will work in the US if they were to travel here and vice versa.

    • Kevin Fitchard

      Hi Raynman,

      The big difference is frequency bands. While most of the T-Mobile networks in Europe use the standard 900/1800 MHz for 2G and 2.1 MHz for 3G the U.S. uses the same bands 850 MHz/1900 MHz for 2G and 3G — all of which the standard iPhone supports. T-Mobile deployed its 3G (though it calls it 4G) network on a different band called AWS (1.7 GHz and 2.1 GHz). Since T-Mobile USA and only a handful of others did so, Apple couldn’t justify the additional expense to support such a small group of carriers. Meanwhile all of the other T-Mobiles use the standard European configuration.

    • Kevin Fitchard

      Oh and as for interoperability across the T-Mobile networks. Quad-band phones should be able to roam and vice versa, but only on 2G. The 3G networks are isolated.

  4. When it comes to antenna’s, I think of Nokias 701 as having pentaband 3G, NFC, FM radio, FM transmitter, WIFI, Bluetooth and GPS to show that a lot can be done about having differing radios in one device. That is what makes the mobile phone much more than just a micro-sized pc. Apple could also put AWS into a revamped Iphone 3GS, 4 or 4S when they put out the Iphone 5. It’s all about money and if the price is right…why not?

    • That is an EXCELLENT point! Apple is obviously very money hungry, and if they were willing to make carrier-specific variants of the iPhone, that would be a big opportunity of revenue for them. I assume most T-Mo customers (myself included) wouldn’t mind getting a revamped 4S as the next iPhone comes out… because I would be happy to finally have an iPhone on T-Mobile, not to mention the 4S is a pretty awesome device. Also considering T-Mo won’t have LTE anytime soon, one of the already released devices would be more than sufficient for their network. I just wish that Apple would think outside the box more and make ideas like this into reality!

    • Kevin Fitchard

      Hi Stuart, Brent,

      I agree with you buys that the iPhone opportunity for T-Mobile will be the launch of the iPhone 5, but I think the device it would get would be the 5 not the 4S or the 4. The reason is Apple will likely have to make multiple variants of the 5 to capture all of the global LTE bands. It might even have to make several North American variants.

      If that’s the case, then Apple has much more an incentive to include AWS HSPA+ support in one of those North American variants because it’s only spreading the additional band/radio costs across the millions of devices in that SKU, not across all iPhones of the world. It might suck to get an LTE iPhone with no access to LTE but a 21 Mbps or 42 Mbps phone on T-Mobile’s network would be so fast I doubt anyone would notice, plus it will have a lot longer battery life. Come to think of it an iPhone 5 on T-Mo’s network might actually be better than the same device on AT&T and VZ.

      • Hey folks,
        There is much misinformation floating around here. VZW and AT&T have gobs of AWS-1 Spectrum in the USA, as well as T-Mobile. AT&T continues to tout its HSPA+ service while building the LTE network out. Frequency bands are standardized to simplify product design and manufacturing (e.g. LTE bands at The same motherboard and chipset are used to support the different band configs in different regions. As for antennas, consider that AWS-1 straddles above and below the PCS band in USA. The antenna design needs to be broadbanded to accommodate the wider spectrum. iPhone 4s specs say: “World phone
        UMTS/HSDPA/HSUPA (850, 900, 1900, 2100 MHz);
        GSM/EDGE (850, 900, 1800, 1900 MHz)
        and CDMA EV-DO Rev. A (800, 1900 MHz)” (the latter is only for phones provided by CDMA service providers). This listing does not include UMTS band IV ( AWS. Since the Euro 2100 MHz band is already in 4S, the antenna(s) already is(are) quite broad (800 to 2100 MHz). AWS therefore is likely to be no-brainer for the next model. Also, LTE does not support voice yet, so a VZW phone that does LTE also has to support CDMA-family technology today. Likewise, an AT&T phone that does LTE also has to support GSM and might as well support UMTS/HSPA in the process. And the technology is not exclusive to a spectrum band. LTE can operate on any band listed in the standard, so the devices should be ready for that (e.g. AT&T and VZW LTE buildout may focus on 700 MHz spectrum at the moment, but over time – or in an area where a carrier lacks 700 MHz – LTE can go on Cellular, PCS or AWS and the LTE capable device should be able to handle it. Ergo, the iPhone technological issue for T-Mobile is when AWS gets on iPhone, not if. HSPA already there. The business issue (making a deal with Apple) is a whole other kettle of fish!

  5. Apple should just buy T-Mobile and convert it to a data only G wireless network. If you want to make a call on “Apple mobile”, you do so over the internet using an Apple “Skype” like service or Skype, Google, etc.
    Having total control over the network means iPhones on the Apple data network can be made to work to Apple’s specifications and not to the capricious whims of telco executives and engineers.

    • Kevin Fitchard

      Hi Mike, the only problem is that operators make most of their money off of voice. If you get rid of the voice of the network and become an all-data operator you no longer have that huge cushion of revenues (with little investment) to fall back upon. That’s a hard business case to make. Just look at the problems Clearwire is having. And as for capricious whims, I don’t think that applies here. If Apple bought T-Mobile it couldn’t change where HSPA+ sits on the electromagnetic spectrum. It would have the exact same problem of handset design.

    • Kevin Fitchard

      Hi WSJ,

      The thing is it’s not just that single dollar. Apple makes a single SKU for the iPhone that means it adds $1 to the cost of each of the hundreds of millions phones it sells across world, just so it can each reach a few million T-Mobile USA customers. That math doesn’t add up unless Apple changes its single SKU model.

  6. arnotron

    As reader from Germany I always wonder about articles about T-Mobile, sounding like the only market this brand operates in was the US. T-Mobile already had the iPhone from the first day, exclusively in Germany.

    • Kevin Fitchard

      Ha! I guess we do tend to forget about the other T-Mobiles, writing from a U.S. perspective. In Europe do regionally focused media always spell out the specific T-Mobile or Vodafone they’re referring to? I know Europe is far less insular in the U.S. so I’m curious.

    • Derek Kerton

      Yeah, but irrelevant to the discussion at hand. T-Mo, in the USA, shares little more than ownership and a brand name with Deutsche’s EU properties. The different spectrum positions assure that divide to be long-term.

    • Derek Kerton

      Which brings up another point not yet mentioned.

      Strategically, from an Apple perspective, what additional addressable market does a customized iPhone for T-mo offer them? Existing T-Mo customers absolutely CAN buy an iPhone…they just need to go to AT&T or VZW to do so…and many have.

      The need to churn a customer to sell them an iPhone is a barrier for an iPhone sale, but Apple needs to consider the strength of this barrier versus the one Kevin discusses in the article, above.

      Sadly for T-Mo, with such an iconic device that customers actively seek, Apple doesn’t need to be on T-Mo to sell phones to T-Mo subscribers.

  7. Andrew J Shepherd

    Hi Kevin…

    Chipset capability should not be the issue. Modern chipsets can (and frequently do) support the gamut of airlink technologies and band classes for both 3GPP and 3GPP2 families. In fact, current iPhone chipsets already support AWS 2100+1700 MHz for W-CDMA band class 4. See this table from Qualcomm by way of AnandTech:

    But an airlink or band class capability in a chipset does not always extend to a handset that utilizes said chipset. As you note, a handset concomitantly requires similarly capable amps and antennas feeding to/from the chipset. If no AWS antenna, then no AWS capability, regardless of chipset.

    Even when chipsets, amps, and antennas are all in place, carrier-manufacturer politics also can play a role. For example, AT&T LTE handsets support AWS for LTE band class 4. As such, those handsets have amps and antennas able to support as well W-CDMA band class 4, yet firmware omits or disables that capability — quite possibly at AT&T’s behest.


    • Kevin Fitchard

      Hey AJ, good to hear from you.

      So what’s you best guess on what LTE band configs will look like after these initial batch of carrier-specific phones have run their course?

      • Andrew J Shepherd

        Kevin, that is a very good question. But any answer that I could offer at this early stage would be, honestly, nothing more than a shot in the dark. What I can say with certainty is that, in order to be compatible with just the LTE networks already in commercial service, any LTE capable iPhone model(s) would need to include support for at least the following LTE band classes:

        North America:
        Lower 700 MHz (band class 12 or 17)
        Upper 700 MHz (band class 13)
        AWS 2100+1700 MHz (band class 4)

        Eurasia and Australasia:
        DCS 1800 MHz (band class 3)
        IMT-E 2500-2600 MHz (band class 7)(TDD band class 38)

        Digital Dividend 800 MHz (band class 20)

        IMT 2100+1900 MHz (band class 1)

        By comparison, the current single inventory iPhone 4S needs to include support for only four W-CDMA band classes.


      • Kevin Fitchard

        Wow, so if Apple wants to cover just its largest markets with LTE in a single phone it would need 8 bands LTE bands supported, and that doesn’t even include the legacy 2G and 3G bands. What fun for Apple designers!

  8. No smartphone, especially not the iphone, has small margins. In fact, compared to most consumer electronics, smartphones have ridiculously large margins. Carriers pay $500 to $650 for the most recent round of high end smartphones, and I doubt if any of them cost more than $250 to make, if that.

    as for engineering costs, if they sell 5M iphones a year on T-Mo, they can easily justify spending a few million on engineering the changes. If Samsung can build a Galaxy Nexus that runs on T-Mo’s network, I think apple, with it’s much higher margins and sales volume, could also.

    • Kevin Fitchard

      All good points, Ken. HTC, Samsung, Nokia, LG and RIM all see the value in making phones for T-Mo. But they make variants for T-Mo’s frequencies. Apple insists on doing a single SKU whenever possible. And that means whatever additional costs go into building AWS into the iPhone are passed onto every customer, not just T-Mo. If it would just build more SKUs it wouldn’t be an issues. But that doesn’t seem to be in Apple’s DNA.

      Keep the comments coming.

      • Are you saying the iphones for VZ/Sprint and ATT are the same design? I don’t think they are, and if they are the same, it would be easier for Apple to make an ATT/T-Mo version and a VZ/Sprint, then to combine all four into one design, as there is much more commonality between ATT and T-Mo. I realize that ATT will be adding LTE, and will want it in all new iphones, but that will only make the T-Mo version less expensive.

        I would bet the only reason they haven’t introduced a T-Mo iphone is they wanted to wait to see if ATT’s acquisition went through. Since it’s dead, they will want coverage on all U.S. networks, as it won’t cost them a whole lot of engineering or product management to get that.

      • Andrew J Shepherd

        Supposedly (though I cannot recall the source, maybe Qualcomm?), it is most feasible for current LTE devices to support maximally two LTE band classes 1 GHz. This, combined with the fragmentation of LTE band classes in the Upper/Lower 700 MHz allocations, not to mention in other spectrum bands around the world, will likely present an engineering challenge for Apple. Thus, if the next iPhone supports LTE, then I do agree that Apple may be forced to return to multiple SKUs.


      • Andrew J Shepherd

        Apparently, the posting system did not like my use of greater than, less than symbols. This is how my truncated post was intended to read: “…it is most feasible for current LTE devices to support maximally two LTE band classes below 1 GHz and three LTE band classes above 1 GHz.”


    • Galaxy Nexus isn’t the best example. There are separate GSM and CDMA models, so it’s not like Samsung built one model that runs of all/most carriers. Additionally, the GSM Galaxy Nexus “runs” on T-Mobile’s network, but is limited to HSPA+ 21 rather than the more desirable HSPA+ 42. Lastly, Apple makes in iPhone that runs on both CDMA/GSM.

      • Kevin Fitchard

        I hear everything you’re saying, Ken, and what you’re describing would be the path that any handset would take. But Apple has always been the exception. Except for its brief trist with Verizon last year, it doesn’t make variants for specific carriers. Like Raymond just pointed out, the CDMA and GSM versions ARE the same device. That may sound crazy to you and me, but it obviously works for Apple.

      • But the iphone that sells on VZ and ATT is not 4G capable, that’s a big deal. They’re not dealing with as many bands. And HSPA+21 won’t be a big deal for Apple, for they don’t even ship a 4G phone now.