Your browser is not supported. Please update it.

10 June 2019

If Amazon wants Boost, it will be to support AWS, not Alexa

One of the concessions T-Mobile USA and Sprint have proposed, to persuade the Department of Justice to approve their merger, is to offload Sprint’s Boost Mobile prepaid subsidiary in tandem with an ‘infrastructure MVNO’ deal. That would presumably involve some spectrum and network assets, laying the foundation for a new mobile player to allay the antitrust regulator’s fears of reduced competition.

Reports indicate that there are several parties interested in Boost, most of them from the prepaid market (see inset), but the most interesting of the rumoured candidates is Amazon AWS. The webscale giant has been increasingly active in wireless connectivity trials, raising the prospect that it may want to roll out its own network.

According to reports by Reuters, Amazon is considering a bid for Boost Mobile – not, presumably, for the brand, which is tiny compared to Amazon’s own, or even the customer base, but for the promised rights to access ‘New T-Mobile’s’ network for six years, and possibly some spectrum too, if that is included in the deal.

Sprint and TMO set out the terms the proposed offering an MVNO in an FCC filing, saying: “New T-Mobile will offer the Boost buyer terms for a six-year wholesale MVNO agreement that will include wholesale rates that will meaningfully improve upon the commercial terms reflected in the most favorable of T-Mobile’s and Sprint’s three largest MVNO agreements.”

Even the hint that Amazon might enter the mobile business sent operator stocks sliding, since the webscaler would have the kind of scale, technological agility and customer relationships to be a serious threat to established telcos.

Some commentators thought a move of this kind would be a way to test out a new business through a lateral acquisition, as with Amazon’s purchase of Whole Foods in the grocery space.

“This would be an excellent vehicle for Amazon to dip its toes into the wireless business,” said analyst Mark Lowenstein of Mobile Ecosystem. “It’s undoubtedly a better network deal than they would get from AT&T or Verizon. Dish is still a wild card and the timeframe too uncertain.” He thinks the motivation would be to provide a strong testing ground for new consumer experiences built around Alexa and other devices created by Amazon’s secretive R&D arm.

But is it really probable that the company wants to be a consumer-facing MVNO or MNO, and does it really need to control a network in order to test future Alexas? It looked set to write new rules for wireless data when it launched its Kindle ereaders, with cellular connectivity bundled in the price – but that disruptive model was not viable for devices with far greater data usage levels. While it made a success of the Kindle Fire, its Kindle Fire smartphone was a flop, and highlighted how Amazon can drive consumption of its retail and content services through addictive subscription offers like Prime, with no need to control the network.

It does not need an MVNO deal to reduce the cost of streaming content when this can be done over-the-top, at the user’s expense. It could want to optimize the user experience to differentiate Prime video from competitors, but in that case it would need more control over the network than most MVNO deals provide – indeed, recent studies have shown that MVNOs are given slower network speeds than their hosts keep for themselves. But acquiring spectrum and building its own network would be an expensive way to deliver marginal gains in user experience, in a world where most of its users probably stream video over WiFi anyway.

It is far more likely that Amazon is interested in wireless connectivity to boost its AWS cloud operation, which has been the main business unit involved to date in experiments with spectrum and wireless technology. Forget the expendable Boost brand and customers – this could be a way to access spectrum, an alternative to AWS’s previous real and reported wireless initiatives, from Globalstar’s TLPS WiFi-like band, to tests in CBRS and WiFi spectrum, to a rumored desire to be the anchor tenant for Dish’s network.

T-Mobile and Sprint have said they would resist concessions that would reduce the deal value significantly, which would presumably include significant spectrum divestments, but a small amount of 2.5 GHz (of which Sprint has plenty) or millimeter wave might be enough to support Amazon in building its own wireless networks for selected industries or applications, enhancing AWS’s rapidly evolving edge compute strategy. It is highly unlikely TMO and Sprint will divest enough spectrum to make AWS a full mobile player, and there is no reason for it to become one. But it could certainly emulate the model that the large cablecos are interested in – combining WiFi, MVNO deals and localized cellular with a local core and edge node, to create a fully controlled sub-net and provide an optimized experience for a particular enterprise, city or hotzone.

This would help AWS improve its position in some enterprises, against the entrenched Microsoft Azure, and would help it deliver an on-premise edge cloud/mobile platform that could achieve two important goals for the main cloud business – help tie customers to AWS, when many are adopting fickle multi-cloud strategies that commoditize the webscalers; and provide a reassuringly local toe in the water for organizations which have not yet taken the plunge into public cloud (a stated objective of AWS’s edge strategy).

Amazon has already indicated this kind of thinking in shared spectrum, most recently CBRS. It partnered with Ruckus Networks, Federated Wireless and others to launch “a fully cloud-native private mobile network solution for developers, ISVs, telecom operators, public sector and enterprises for quick deployment of Industrial IoT applications, such as real time surveillance, smart meters and worker safety monitoring”. Providing optimized connectivity for emerging IoT and enterprise applications would help AWS steal a march on the MNOs in some sectors where they have failed to deliver strong wireless performance, and this has been the focus of its reported dalliances with Dish, which is building an NB-IoT network in some of its patchwork of airwaves.

AWS has trialled various emerging spectrum bands. In 2013, it worked with satellite provider Globalstar, which owns S-band spectrum (2483.5-2495 MHz) adjacent to the unlicensed 2.4 GHz ISM band, used by WiFi and other radios. The two companies trialled TLPS (terrestrial low power service), with a view to running a private network, using WiFi-like technology. That would have been able to support WiFi devices while offering partners a more secure and quality-controlled spectrum environment than the 2.4 GHz wild west.

The plan foundered on FCC and mobile industry opposition, and fears of interference with WiFi, but it was an early indicator of the desire of some non-MNOs, such as Amazon, to find a network they can control more effectively than an MVNO agreement, but not actually have to build out.

Over the past year, there have been persistent reports that Amazon was interested in being an anchor tenant, and co-investor, for Dish’s NB-IoT network, which the pay-TV provider is already promoting as ‘5G’.

The thinking behind a tie-up between Dish and Amazon is that the retailer would help fund the build-out of a network in Dish’s spectrum. For Amazon, that would bring it a mobile network which could be optimized for its own ends, from internal IoT usage to AWS services to supporting mobile Prime services. This would come at far lower cost than seeking its own spectrum, but with better control than an MVNO or WiFi approach. For Dish, the investment and the anchor tenant would make the economics work at last, and it could also launch mobile options to its TV customers in bundles, to make it more competitive with AT&T’s DirecTV and with cable providers.

One way in which such a venture could deliver more than the sum of its parts would be in the IoT. Dish already plans an NB-IoT roll-out, and while this looks largely designed to address FCC build-out demands associated with some of its spectrum, it could be expanded in scope to provide the first US network fully optimized for machine-to-machine purposes, rather than bolted onto a mobile broadband platform and business model.

The resulting network could be used by Amazon, as the anchor customer, for its own logistics purposes (it is an increasingly heavy user of wireless delivery and tracking, including drones); and to offer IoT services to enterprise AWS customers. AWS could layer new services onto a wireless network over which it had significant control. The recently launched AWS Greengrass edge cloud platform could be an effective way to address low latency IoT requirements, for instance, if integrated into the mobile network.

And at AWS’s annual developer event, re:Invent, last autumn, there was a demonstration of DeepLens cameras connected to a private LTE network running in CBRS shared spectrum. This highlighted how non-MNO spectrum will allow alternative providers to seize some of the rising enterprise cellular business from the traditional operators; and how interested Amazon is in that development.

Like other enterprise providers, AWS could use shared bands, like the general access element of the USA’s CBRS spectrum, to move more deeply into connectivity without having to buy licences or pay hefty fees to MNOs. The demo, at AWS’s Las Vegas event was clearly designed to show enterprises how they could use CBRS for their own private mobile purposes, with the help of their cloud provider rather than their MNO.

About 100 developers were invited to take part, and each used a DeepLens device, built by Amazon with facial and image recognition capabilities. The devices linked to modems over the 3.5 GHz CBRS band, enabling all the cameras to exchange data within the conference hall.

Federated Wireless, one of the providers of Spectrum Access Systems (SAS) to manage the various tiers of access to CBRS frequencies, supplied the spectrum controller, while AWS contributed the cloud platform, and Amazon a range of services.

The demonstration also used the BubbleCloud local packet core from Athonet, one of the start-ups being incubated by Orange under the Facebook-driven Telecom Infra Project. In this set-up, BubbleCloud enabled localized LTE networks connected and controlled from the AWS cloud, while the LTE RAN equipment itself came from Ruckus Networks.