The Signal #47: You’re holding your 5G handset wrong

from our 9/7/2019 newsletter

Friends – our last post discussed how broad 5G access (in particular, for mm-wave 5G) will likely be driven by mundane but important operational concerns like pole access. In came the question (thanks, Eric) – what about T-Mobile? Would T-Mobile need to get on every lamppost in San Francisco? (Note T-Mobile already has a lot of DAS poles in San Francisco via partner Crown Castle.)

Good question. The answer comes down to how T-Mobile applies its 600 MHz holdings.  

A little background: T-Mobile acquired 600 MHz in the 2017 600 MHz auction, aka the Broadcast Incentive Auction. That was the second auction of former TV spectrum, following the 700 MHz auction in 2008

For spectrum nerds out there, here’s the 600 MHz band plan.

TV to the left, cellular to the right

Now, in this plan, uplink is on higher spectrum than downlink. In cellular, when using paired spectrum, usually uplink is lower and downlink is higher. Handsets are battery powered and lower spectrum travels further.  Thus, having signals sent from the handset up to cell tower on the lower band puts less burden on the battery, which makes for a happier consumer.  In the 600 MHz case, putting uplink on the higher frequency closer to 700 MHz may be for antenna length reasons. Essentially, the crux of this post.

T-Mobile was the biggest winner at the 600 MHz auction.  The propagation characteristics of 600 MHz and their narrower bandwidth would lend it to use as what operators would call a macro overlay, which provides broad coverage of a city without needing a densified network.  As spectrum, it is cost-efficient from a site acquisition perspective.  T-Mobile has fielded 600 MHz for LTE – if you are a T-Mobile customer, check if your LTE supports Band 71. Here’s a list of compatible handsets. If it does, it is capable of using 600 MHz spectrum.  

Now, if used for 5G,  the narrower bandwidth of 600 MHz channels should make it, well, 5G Lite.  Assuming ~23 b/s/hz for 5G, at 5 MHz bandwidth, then theoretical maximum speeds (without factoring in overhead, or interference from a human hand, etc) would be ~115 Mbps (not accounting for any MIMO magic), and significantly slower in reality. Fast for cellular, but not a fiber substitute. 

Recall that when 600 MHz was used for TV, it was received through a loop antenna.  Which makes me wonder if 600 MHz phones would be better off with an external (extensible) antenna, much as early cellphones had

back to the future

Which reminds me of one of the more amusing moments of the early smartphone era – in 2010, Steve Jobs telling iPhone 4 owners they were holding their phones wrong.  Really, this happened, and only Steve Jobs could get away with that. As a thought exercise, imagine Tim Cook saying something similar.  Or saying, hey, just buy a pair of AirPods. 

Here’s a capture from the iPhone 4 keynote.

Look, magic!

And a great breakdown from Anandtech as to whether the iPhone 4 antenna was actually better than the 3Gs.  Eventually, Apple sent iPhone 4 owners plastic bumpers, which put a buffer between the phone and the bag of RF-impeding water that is a human.

I’m somewhat surprised antenna reception issues haven’t come up with T-Mobile 600 MHz customers (Reddit thread here. Generally, Reddit or T-Mobile support posts on 600 MHz seem to be about limited coverage. With the repacking of TV channels likely happening on an MSA by MSA basis, this is to be expected).    A wavelength calculator indicates that half-wavelength at 663 MHz would be about 9 inches. Longer than most smartphones, but not longer than a tablet.  

T-Mobile has tested 5G at 600 MHz and is certainly talking about it. (The Samsung S10 5G supports Band 71, but for LTE, not 5G NR.)  But, at some point, will John Legere tell consumers they’re holding their 600 MHz 5G phones wrong? Or just send them an externally attachable antenna?

Excuse me,  I’m going to take this call next to a window.

(h/t Jay Goldberg)

The Signal #44: 5G looks like the WISP business

From our February 17, 2019 newsletter


Let’s get to it. This year, we’re focusing in on two subjects: Future Japan, and 5G and security.  

First, Future Japan. Here we will introduce companies exemplifying the road ahead for Japan, after the 2020 Tokyo Olympics, after decades as asset-centered, asset-exporting economy.  If you were to think of one company synonymous with Japan Inc now, that would probably be Toyota. Future Japan? It may be Recruit. It may be FANUC. It may be Fujifilm/Cellular Dynamics. It may be a startup like Spacemarket. It could even be Airbnb, as Japan focuses more and more on hospitality, tourism and asset sharing. (Airbnb is right in the center of that Venn diagram.) Future Japan, potentially, looks like a hybrid of Germany (manufacturing and suppliers; medicine), France (food and culture; nuclear), and the UK (idiosyncratic, historical, select sectors of excellence).

Future economy doesn’t mean new. Recruit is close to 60 years old. But as a firm in the recruiting (full time and temp) business that’s investing in AI and infrastructure, it embodies a human capital centered company investing into an era of a shrinking workforce.  Nippon Life is another – not only is its customer base shrinking, but so too is its actuary base. This should be noteworthy for other human capital firms in other mature economies. Japan won’t be alone in managing an aging population – South Korea, and China, among others, are facing or will face similar challenges.

5G and security? These days, a “5G security” Google alert will produce a lot of geopolitical coverage – namely, whether Huawei represents a genuine cybersecurity risk. Note that Huawei is the #1 wireless infrastructure provider in the world. This is not a startup. Blocking Huawei from future 5G deployments is not a trivial act. Especially since 5G networks generally won’t exist in isolation – rather, they co-exist with earlier generations during a decades-long migration process.  (Imagine apartment dwellers slowly being moved upstairs, but the landlord needing to keep the lower floors neat and functional and revenue-creating, and some lower floor dwellers saying they’ve got rent control and like their current apartment just fine, thank you. A parking meter with a 2G/3G modem counts as one of those lower floor dwellers.) So, an incumbent Huawei customer, of 4G infrastructure, that chooses to not adopt Huawei for 5G will lose some operational efficiencies, and may even have to rethink how its 4G network interoperates with other generations. How important this is…is not really an economic question.

But, it’s likely that Ericsson and Nokia, and perhaps Samsung, will be beneficiaries of Huawei’s straits. (The network infrastructure market has quietly become a highly concentrated market. I pointed this out to a Haas economist, whose eyes boggled and marveled how this had escaped notice. There’s virtue in being quietly valuable and not competing with your customers. One of my students asked why Ericsson, featured in a case we discussed, didn’t launch mobile service themselves. The answer – they have hundreds, if not thousands, of mobile operators they can serve, so why compete with their customers?)

 I’ve written about receiving grant funding to assess the security implications of 5G networks, from the UC-Berkeley Center for Long-Term Cybersecurity. At the time I wasn’t thinking about geopolitics *at all*. Rather, there are some features in the specification that leapt out as me as potentially opening new risk surfaces, such as network slicing. Also, I suspected 5G would lead to more heterogeneous service models, more akin to those in IT for the enterprise. And that heterogeneity could lead to added risk.  Here’s a capture from my recent class on network operators. The lower left (“definite differences”) is what I know to be true; the lower right (“potential differences”) is what I think may come true.

5G known knowns and potential impacts

Let’s go back in the wayback machine for a moment.

4G rollout began in earnest in 2010. (Sprint got off to an early start in 2008 with Wimax. But, an undercapitalized, struggling Sprint was not in a position to capitalize on first mover advantage. Nor was the Wimax ecosystem as developed as that for traditional cellular. So, we’ll take 2010 as our general 4G launchpoint.)  Verizon went on the air in December 2010. DoCoMo did in Japan that same month. Teliasonera launched in Sweden in 2009 and Finland in 2010.

I remember raising a glass in December 2010 with two nervous Verizon employees, one of whom was in a network role, at a certain venture fund’s holiday party. It was the night before launch.  The launch date was just the commercial launch date – the network was “on” already. Still, they were, rightfully, full of jitters.  What would happen when consumers started using it? (“Here, have some holiday champagne? What better day to be hungover?”)

With the benefit of hindsight we can see Verizon did several things *right* with the launch. They rolled their own CDN, rather than using, say, Akamai’s. They invested into the venture ecosystem by partnering closely with several venture firms that in turn invested into technology spheres of interest (network technologies and batteries among them). They started with low(er) band (700 MHz) and laddered on higher bands (AWS) to gap-fill urban areas later.  (I don’t know if that was a strategic plan or opportunistic / reactive.) Perhaps most importantly, they got out there first, visibly, which really catapulted the firm forward vis-à-vis domestic competitors. My own image of Verizon to that point was the firm that was pushing de-featured flipphones (no Bluetooth!) that played MIDI ring tones. Yawn. That image was transformed. And it was in a network-centric way that was consistent with VZ’s brand positioning to date.

Here’s some throwback marketing. There’s honor and profit in running a reliable, fat, pipe.

4G meant a fatter pipe

That was a real ad.

This isn’t to cheerlead for Verizon. But rather to say network transitions present an opportunity for network operators to recast themselves.

To consumers, what 4G did was to make smartphones better. Recall the first iPhone, in 2007, was a 2G EDGE phone on AT&T, due to concern about lack of robust 3G coverage. (Seriously.)  Smartphones were already on the uptake by the time 2010 came around, and, well, they got better. The chief differentiator, when I got my first 4G phone (an HTC Android phone), was that file downloads took less time, and I experienced much less latency. No fireworks, but I could see and feel a difference.

So, let’s apply that logic to 5G. If 5G, now on the air on a limited basis, gets *real* in 2020, what is the 2017 or 2018 service or device launch that will get dramatically better?  


I see video calling around me every day and it’s not hard to think that would get better. In terms of content, mobile gaming could get console-level fun.  Verizon’s recent edtech challenge may provide some hints. But, I would argue there’s more ambiguity this time around. Chatham-house rule conversations with operators – and fiber operators – support that. The question I asked above – what’s the current service that gets dramatically better with 5G? – generally gets some head-scratching in response.

But – that may because of greater heterogeneity in service models. And for that reason, operators seem focused on building reference cases, as the Verizon challenge above illustrates. A European operator shared a manufacturing use case.  In the US, we see it in how Verizon and AT&T are test-marketing 5G. Verizon is positioning it as a home broadband substitute – essentially, fixed wireless to home CPE, which is (perhaps ironically) similar to Clearwire’s Wimax home broadband service before being rolled up into Sprint.  Focusing on fixed wireless reduces the number of variables that need tweaking. Mobility, handoff and roaming aren’t issues. Home coverage (in-building propagation) is. And making sure that content *feels* better, somehow, even if displayed on a fixed PC or TV. So getting access to lightpoles and street furniture, and lining up content providers, become operational priorities. Still, this is the time-honored (W)ISP business.  

Meanwhile, AT&T has launched mobile service, in a way – with a 5G mobile hotspot. It’s here I’ll point out AT&T’s CORD initiative and say that I hope they tie this to 5G rollout. Pushing data centers – 1000s, i.e., one for every legacy central office, not several or maybe 8 or 10 nationwide- close to the user could enable a variety of latency-sensitive services, certainly video/VR among them, or perhaps city infrastructures. I won’t say cars. I’m not a believer. Cars are national or global creatures and 5G rollout is too nascent. (VERY local fleets – maybe.) But a smart (heh) city could harness local 5G in meaningful ways. And AT&T’s CORD strategy seems very aligned with this. Could a city have a more credible wireless network supplier than the former Ma Bell, with the access to the required real estate and light poles?  The benefits of incumbency are large.

Back to the enterprise IT analogy. The enterprise IT market has relatively consolidated equipment suppliers sold through a host of VARs.  (Hey, it’s the channel! Still with us after all these years.) Let’s apply this to 5G for a moment. A national or even regional 5G network, operated by a telco, resold by a VAR to businesses or government, with varying tiers of service and heterogeneous security needs and models. Can a wholesale network operator push all that segmentation onto VARs?  In this model, the telco is a wholesale access provider. In traditional enterprise, this model is mature and established. Thus, I wonder if wholesale 5G to the building or enterprise is another chance for traditional enterprise equipment and security providers to propagate themselves.

The Signal #38: how to start an innovation cluster?

From our January 20th newsletter. (Want to read them in real-time?  Please sign up!)


We often show this diagram in presentations when talking about how Silicon Valley functions. Credit where credit is due: we developed this based on AnnaLee Saxenian’Regional Advantage, which compared Silicon Valley and Boston / Route 128.


Silicon Valley is already a functioning ecosystem. Describing how it works is a matter of observation. But what if you’re starting from scratch?  This is a question many city and regional planners have asked. If you think of Silicon Valley as one big self-sustaining network effect (people come because of what’s there, and therefore what’s there gets better, and therefore more people and capital come), how do you create network effects in, say, Columbus, Ohio? Detroit? Fukuoka?  Of the above five elements, which should come first?

Jon recently visited San Diego and Seattle, in addition to sites in San Francisco, Berkeley and Silicon Valley. The goal: to visit innovation hubs and also examples of urban redevelopment. Naturally, these two themes are linked. As legacy industries are shed, new green shoots emerge if/when they get exposure to the sun. Similarly, if redevelopment can attract companies that attract people, or better yet, attract people who found companies, and their friends, a cluster can be nurtured.

South Lake Union in Seattle, home to Amazon for more than a decade, is an example of redevelopment that was transformative for the host city.  Of course, Amazon, Microsoft, T-Mobile, and Boeing were already in the Seattle area when SLU was redeveloped by Vulcan Realty. The talent pool, and several pillar employers, were already there. The SLU example isn’t lost on other cities; hence the flood of offers to host Amazon HQ2.

If starting more or less from zero, the usual answer of where to start is universities, which gets us into the typical Silicon Valley genesis story of Frederick Terman of Stanford. More generally, universities attract bright students and faculty, who are likely loyal to a place and therefore more inclined to remain than out-of-towners may be to moving there.  (Both remaining and moving to are, of course, contingent on having employment opportunities.) And if bright people are there creating jobs, then capital logically follows. And indeed, Silicon Fen is based around Cambridge and Boston’s community around MIT and Pittsburgh’s around CMU.

Can the opposite happen? Where a company comes first, and a cluster is founded?  The San Diego telecom cluster begat by Qualcomm and its antecedent, Linkabit, is an example. The Linkabit case also doubles as an example of how new companies can mature their technology through government funding.  The Linkabit diaspora is shown in this diagram, created by former Linkabit executive Martha Dennis.


Note that Linkabit was a NASA vendor. (San Diego also has a substantive military presence.)  The multiplier on those government funds, in terms of local economic contribution, is likely immense.

More to come on this subject.

– Team Blue Field

The Signal #35: IT and the Channel

From our November 16, 2017 newsletter.  (Want to read them in real-time?  Please sign up!)

Working in tech, one often hears about “the channel”. As in, we need to develop our channel partners.  Or, we’re dealing with some channel conflict issues. But generally, the channel is an undercovered  area of tech. Equity analysts don’t make their name covering the channel. Tech writers certainly don’t. So here we are to shed some light.

Recently we have done extensive work for a client bringing a networking product to market.  The Internet has radically changed distribution for consumer electronics, but distribution for enterprise electronics is in a much more intermediate stage of change. For any company building a product – hardware or software – for enterprise customers, developing a robust channel strategy can be a make or break proposition.

Broadly speaking, sales through the channel are indirect sales, or sales to some third party who the re-sells your product to the actual end-user. For startups or companies entering a new market indirect sales can be crucial, at the very least as an intermediary stage prior to building out a large (and expensive) direct salesforce. In essence, the startup is borrowing the relationships of the channel partner to address customers it couldn’t normally address as a startup.  In theory, the channel partner, end customer and startup provider all should benefit.

The channel covers a huge variety of partners. These can range from bare-bones distributors who take a point or two for providing basic warehousing and logistics to full-blown, active partnerships handling all sales and customer education. Companies may have more than one channel tier, as shown below.

IT channelMatrix this by territory, market vertical and target buyer size and you can start to see how complex the channel can be.

For companies selling to the enterprise, it is important to think about a channel strategy early. When launching in a new field, companies will need to handle some sales directly to develop demand, get product feedback and build credibility. But soon after that initial phase, the focus will need to shift to broadening customer reach and revenue generation. This can mean deciding on a channel strategy suitable to your end-market. Is your product bought by a large number of small, local customers? This will likely dictate a multi-tier approach with some combination of distributors and re-sellers. Selling large tickets to a small number of customers? This lends itself to partnerships with a few key partners – maybe a white label deal with a larger OEM selling a broader bundle, or a focus on a small number of specialty integrators.

For example selling to schools and hospitals requires work with a huge number of locally-connected but small integrators. In the hospitality market (hotels, casinos, convention centers, etc), there are a few dozen regional players in the US with reach in this market.  (Side note: the hospitality IT market is home to some great inelasticities – broadband or screen needs during events, for example. Then there’s the power traveler who treats the hotel room as a remote office. Who doesn’t want to pay for Wi-Fi as a line item, but won’t stay at a hotel that doesn’t have good Wi-Fi. Leading to customer feedback like this.) CrappyWifi






It is important to note that the channel itself is in flux. Distributors of IT products are searching for their role in the world of the Internet and  cloud computing. Looking at the networking market, on one end of the spectrum are companies like Ubiquiti Networks which has no salesforce (and has software margins). At the other end are majors like Cisco and Dell/EMC. Both have massive salesforces, but still do much of their revenue through the channel. In many cases, these large incumbents have sewed up distributors through 20+ years of certification, incentive programs and relationships. These are barriers to entry, but these relationships offer leverage points to new entrants as well. Many distribution partners look for alternatives to their large suppliers.

Enterprise IT product companies get a beachhead with their first product, which may not be profitable by itself. The second or third product, though, is where profitability can be achieved, especially with a sound channel strategy.  (And we, of course, are happy to help.)


This week’s overheard: Who will fill the gas tanks for the autonomous vehicles?