369 comments

[ 3.6 ms ] story [ 330 ms ] thread
Super excited about the new edge usecases this could potentially enable
What's the business case for a service like this? Is there anyone on HN that would order this for example?
If only the linked-to page had a section clearly headed "Use cases"

/s

I completely missed that
There's use cases further down the page. A fair amount of companies have existing applications that are on-prem and could be made to work with mobile devices this way without having to expose endpoints to the internet. Things like mobile barcode scanners, iot sensors, roving employees with tablets, etc.

Edit: Though there are other options: https://news.ycombinator.com/item?id=29395370

There's a ton of research applications with small sensors.
They outline a few cool business cases. I imagine it could be used by mining corps who are often in the middle of no-where. Set up your own cellular tower and give everyone Amazon branded sim cards and it will be a hell of a lot better than Wi-Fi in specific buildings and/or radio only communications.
I would imagine that Amazon is already using this service internally to support their Amazon.com marketplace effort. If I were to take a stab at Amazon's internal use case for a private 5g network, I would bet they are using it to manage the communications of the Amazon Logistics applications.

Amazon has come to not be held hostage and rely on outside companies for services all throughout their vertical including Fulfillment, Cloud Computing, and Logistics and Delivery. They have abstracted out their core operational dependencies into their own service offerings, so why not a private 5g network next?

Private LTE/5g is huge in the energy sector. You'll find it deployed for various IoT and end user devices at oil fields, offshore rigs, and even in parallel to the major carriers in large metro areas. The latter is fairly common with energy utility providers. One of the US's largest energy suppliers in the South operates what is probably the largest private LTE deployment in North America, using their own transmission towers to cover a majority of two states in the US Southeast. In addition to using this to support their own ops, they lease access out to other businesses.

It's also increasingly common to see private LTE/5g deployed to support municipal government operations.

They talk about it on their webpage, but I've heard this described as the ability to offer an SLA on a wireless network. This means that rather than using hard wired ethernet, company's can instead use a 5G network under their control and offer equivalent level of reliability.
anything big enough that connecting it using wifi might be challenging. E.g. Port, factories, airports, ...
One use-case is that LTE/5G are becoming increasingly attractive as a replacement for land-mobile radio. It used to be sort of common for major cities to have commercial MotoTRBO/OpenSky/iDEN networks for all kinds of business users that wanted LMR without having to pay to install their own equipment. Most of them died out due to stagnation of the technology and competition from cellular providers. There's a bit of excitement that that business sector could be coming back, as cellular equipment gets less expensive and increased data-centric usage has made quality of service on the mainstream cellular networks much more variable.

Almost all major cities operate a private LTE network for city agency use, for example. But for various reasons it's mostly been out of the reach of private ventures. This could be one piece of changing that (5G brings a number of the other pieces).

Intriguing - but the first three questions that immediately came to mind didn't seem to have an immediate answer. (1) What range/power, (2) Pricing (3) Density/Number of connections? I'd love to see a network architecture that shows how they hop from antenna to antenna as well, and whether it's guaranteed with no interruption to service.
This is fantastic, I wonder what the pricing will be. Cellular stuff is always expensive it seems.
Surely the problem is access to spectrum, or is the idea to rely on unused spectrum bands reserved for local experimental use?
With 5G, many countries have reserved space and/or licensing mechanisms to get spectrum access for small local deployments.
Which other countries have such free spectrum ?
There is a ton of bandwidth up in the millimeter wave area if line of sight is not too obstructed. One whole GHz is nothing while with legacy WiFi it could be almost half the actual carrier frequency.
(comment deleted)
In the US, the FCC has specifically allocated 3.5ghz for open use[0]; paying for a license just gives you priority access to the spectrum. Use of the band requires checking in with a server to lease spectrum at a granularity of about 4 minutes. This is akin to if your Wi-Fi router could ask the FCC to give it a channel known to not have any users within a certain radius of itself.

[0] https://en.wikipedia.org/wiki/Citizens_Broadband_Radio_Servi...

If I want to deploy a custom 5G base station on this band 38, what hardware do I need? Are there free software solutions?
It's Band 48 and there are 321 FCC approved devices as of today
Curious if anyone is familiar: if you could purchase these for your home and all your devices that have wifi chips also have 5g chips, how do you choose between 5g and wifi for your home? Would 5g make sense in wifi-like deployments? Any reason this tech hasn’t merged?
I'd love to set this up for my rural community. I've got a good connection but many people have no cell service, no dsl lines, no cable...literally nothing. Until something like starlink is ubiquitous i feel like this could go a long way to solving their problems.
Came here to say exactly this. Out in rural Canada (Ontario) it is extremely difficult to find reliable, fast, unlimited access.

With Starlink delaying a good deal of preorders, something like this could be great.

It isn't replacement to Starlink because you need a backhaul. You can however take one Starlink and use it to distribute among others. Bottleneck is the backhaul
Can you email me haseeb@haseebawan.com - I would love to talk about it since we're working on solving this
I don't think this solves the last-mile problem. You still have to provide (get) internet connectivity; Amazon just relays that landline to 5G.

And in rural areas, 5G probably doesn't give you enough range anyhow. Have you considered mesh revenue-share networks like Althea (https://www.althea.net/)? There's a nonprofit one too operating in several cities, but I can't remember what it's called.

Starlink could be that backhaul. Future terminals will have 1Gbps.
You don't even need that much bandwidth. I had no problem doing LTE over WiFi with a 1Mbit DSL line, with like 256k up on a good day.
Man, all that work to get worse voice quality than the 1900s... on a good day.
5G can operate at lower frequencies, from my notes I do not remember where I copied it from.

"In quick summary, the bands work as follows in the real world. One low band (600-700MHz) tower can cover hundreds of square miles with 5G service that ranges in speed from 30 to 250 megabits per second (Mbps). A mid band (2.5/3.5GHz) tower covers a several-mile radius with 5G that currently ranges from 100 to 900Mbps. Lastly, a high band (millimeter wave/24-39GHz) tower covers a one-mile or lower radius while delivering roughly 1-3Gbps speeds. Each of these tiers will improve in performance over time."

My understanding is that 5g can operate at the same or similar frequencies as 4g, and actually then has slightly better range and throughout than 4g.
Doesn't look like this will give you phone/SMS service.

You have to put in their SIM, so unless you have a dual SIM phone that can intelligently roam, doesn't seem like the right solution.

It can be used in a modem to provide home internet. Also dual SIM phone are easier to find than laying cables.
Do they've the requirement for super high speed internet or they just want a decent internet ?
Open Source software stack to run private LTE/5G with SDR https://github.com/srsran/srsRAN
For folks who have actually set up srsRAN for anything besides a trivial configuration w/ one eNB and manual management of most things (and even then), the value-add of having Amazon worry about most of that stuff is clear. It's, ah, ... complicated.

This is a bit like responding to, say, the original announcement of EC2 with a link to download a Linux distribution.

I wonder if this is a product that was built on top of something they needed to use internally. Makes me wonder how Amazon is using this technology for themselves... Anyone care to speculate?
I see a potential use of it from IoT devices. An eg. a tractor wants to communicate with a central server for commands.
You describe a great use case for LoRaWAN:

  https://lora-alliance.org/about-lorawan/
I fear what this means in larger scale. All Amazon manufactured physical IoT and hardware could have sidechannel in the future to escape network isolation. They already have this "feature" which expands your home network for neighbor Amazon devices if they need Internet access. And default setting is "ON", not off. How many consumer is aware of that?

https://www.amazon.com/Amazon-Sidewalk/b/?node=21328123011

This capability is considered in the design of 5G, see NB-IoT and LTE-M
It's worth noting that the behavior of Amazon with its commodity consumer products has been notably different than AWS with its customers. The stain on reputation if AWS were to, say, mine data from private S3 buckets would be very hard to remove.
Why would you buy a consumer IoT product and not connect it to the internet? If you’re the kind of person who fears that, you’re probably not the person who uses something like Alexa and smart home stuff.
If the device has not easily removable battery, it starts to be impossible to tell when it collects the data and shares it. Many devices are useful in the local network only, but still they want all-time internet access without real need. Using direct 5G links make it also harder to filter traffic. I don't know, I just don't like the idea that all information must be collected by any means.
> Many devices are useful in the local network only

no amazon-made device is local only.

> I just don't like the idea that all information must be collected by any means.

I agree with your sentiment, but most devices don't really have access to much data of concern. I'd be more worried 5g bullshit is used in screens that can send ads over a smart bulb or something.

I would guess it's more aimed at use in large factories, warehouses and yards, where WiFi is not going to be practical.

I did a stint as a (software) architect for a large Norwegian engineering company, and at the time they were looking at getting a private 4G network setup, as their facilities were absolutely huge. I did a little research, and quite a few mobile operators actually offer private 4G networks for exactly this use case.

(comment deleted)
Could you explain what makes this tech better than Wifi? Does it have better range/deal with interference better?

My impression was that the high speed profile of 5G was basically the same as that of Wifi, with exactly the same issues. Am I wrong/is it better in lower speed modes?

Here are the key differences between 5G and Wifi: 1. Dedicated vs shared spectrum. Though all big countries have shared spectrum initiatives for 5G too but it is still not a free for all like Wifi. So interference-wise 5G might be better for some use cases. Have heard about that in several shipping ports where Private 5G is deployed 2. Security. Due to the usage of SIM but Wifi security is good too 3. Range - though most of 5G is in comparable frequency ranges with Wifi, there is a huge range of powers at which 5G base stations can transmit so range is possibly larger for 5G

But it all depends on use-case and there is no clear winner for all situations.

You could imagine them using it in their warehouses. I imagine getting good coverage with wifi in a huge warehouse could be pretty expensive, maybe 4G/5G works better?
No internal use. This is a hype driven white labeling.
One of the podcasts I listen to had AT&T as their sponsor and they would talk about using IoT to make factories and warehouses "smart", so it's possible amazon uses this in their warehouses instead of wifi to connect to the bots and hand terminals.
I feel like this has come off the back of Amazon creating their own 5G for their warehouses and robots. Can any Amazon employees confirm this?
Can confirm there’s no internal use. It’s a fresh project to capitalize on 5g hype.

They’re working with a partner for hardware but redoing their software. So it’s a pretty low-touch white label job.

They have internal use planned but it was started explicitly for sale to external customers not internal.

I wonder which smart TV provider will be the first to use this instead of home wifi where people can disable or black hole connectivity.
Some smart TVs already ship with sim cards. It just requires a deal with AT&T/Tmo/etc.
Don't think that's the use case. how would AWS deploy a base station to my home?

But you have a real point, and AWS already solved that with the 3G enabled Kindles by having carrier agreements. Why not doing that with Smart TVs?

> But you have a real point, and AWS already solved that with the 3G enabled Kindles by having carrier agreements.

Amazon regretted that pretty soon after they did it, people hacked their kindles to be hotspots and it became an arms race amazon didn't wanna play.

Was that a big problem? I thought it was a single-intern sized problem.
Yeah tbh not hard to search for high bandwidth users abusing system and dealing with it on a case by case basis
5G E2E Slicing will prevent such scenarios. The SIM will be bound to a network slice and restrictions will be dictated by the carrier.
> Don't think that's the use case. how would AWS deploy a base station to my home?

Unless I'm misunderstanding how this works, it wouldn't need a base station in your home, just a commodity cellular modem with a network-specific SIM, and an uplink tower somewhere within range.

yup, the only thing in the past that was keeping companies from popping a sim into every product to report back home (above a certain price) was that there was nothing stopping you from pulling the sim and just using their plan in other devices, at least for a short while until the company realized what was up (if they cared to check (they usually didnt))

now they have complete control over the end-to-end, and can cheaply provision sims that only talk to their local tower for example, and reject non-company provisioned IMEI (if they need it anymore?) etc

working on building blocking for this :)

(comment deleted)
> how would AWS deploy a base station to my home?

The TV company will deploy base station(s) in my town. Embed SIM cards in TVs they ship. No need to connect to my home WiFi to send personalised data back to their servers, or to upgrade firmware etc.,

DirecTV already does this for their 'wireless' offering - you get a huge LTE modem and a bunch of receivers with sim cards in them.
It may interest you to know ATSC 3.0 has specifications for a "return path" to the broadcast station in it.
What does a smart TV have to do with a private 5G network? People just upvote any meaningless comment on this site as long as it sounds mildly controversial.
Smart TVs scan what's on the screen to get data on what people are watching and sell that data to anyone who wants it

The idea is to make sure they can keep that data flowing even if someone intentionally disconnects the TV from the internet

So they can (and sometimes already do) insert a SIM card in the TV and use any of the consumer 4G/LTE/5G networks out there which your phone connects to. Why are people hypothesizing that Vizio will ship a $250K 5G base station to your house?
because if setting up private 5G networks becomes cheaper and open to any business then the chance TV manufacturers consider putting 5G base near enough your house.
But it’s cheaper and easier to just embed a normal 4g modem and SIM card
Every device will have 5G instead of Wifi. Remember Wifi dongles ?
"Please don't comment about the voting on comments. It never does any good, and it makes boring reading." - The Hacker News Guidelines (https://news.ycombinator.com/newsguidelines.html)

Given that there's an extremely simple and straightforward answer to your question (smart TV manufacturers sell data obtained from spying on users, and they need a way to get that data from the TV to their servers), I would encourage you to give some thought before making such an empty and vapid comment about voting habits on HN.

I don’t think enough people use pihole for smart TV manufacturers to care, and encrypted DNS breaks that anyway so setting up a 5G base station in your house would be overkill
This is a platform AWS is providing to make it easier for those who have private 5G networks to run/maintain those networks.

You seem to be interpreting this as "AWS is going to put 5G towers everywhere and smart TVs are going to connect to them to send data they collect!". That's not what this is at all.

This is one of the many reasons it's beneficial to read the actual article/post and comprehend the information, rather than reflexively reacting to keywords you notice in the title.

Exactly, it's more of a b2c thing. Actually in one of my previous gigs (at a energy company that had oil rigs in very remote areas), they wanted to quicken the provisioning of 5g to some of the places they operate in.

AWS was one of the companies they were negotiating with - I never realized they had not yet announced this service.

They don't need this for that. Instead they are using rings and echo devices that allow devices to connect and send small data out. You can't black hole them without black holing the ring/echos (making them useless). You can disable this feature but it's opt-out, and it may just connect to your neighbor's instead.

I dont' remember the exact feature name though.

I believe you're referring to Amazon Sidewalk. It's more for their own devices like cameras and not for TV ad tracking (at least for now.)

https://www.amazon.com/Amazon-Sidewalk/

Yeah that's it, but on the blog post at https://www.aboutamazon.com/news/devices/introducing-amazon-... they say

> For device makers, we plan to publish protocols that any manufacturer can use to build reliable, low-power, low-cost devices that benefit from access to long-range, low-bandwidth wireless connections. In the meantime, you can sign up to be notified when more information is available.

So the intention is definitely there that device manufacturers can pay to delvier data over the network.

I immediately thought of alarm companies like ADT or Vivant who I believe currently partner with cellular providers for access. How much cheaper would it be to swap to 5G as a service versus whatever their current cost model is.
One response to this problem is to perform mild electronics surgery and install fine copper mesh around anything that looks like it moves electrons as part of its normal function. It really doesn't take a whole lot to completely fuck up an RF signal.

Alternatively, you could maybe do the same to your drywall if you are looking at new home construction... If every room is effectively a faraday cage, you are back in control over what can talk to what on a much more granular level. This clearly creates challenges for your own wireless/mobile signals, but presumably you also have the ability to hardwire additional access points if you are going to this extent.

Spectrum questions aside (which obviously one of the biggest one), could this enable running Helium 5G on AWS?

-- Edit: I don't own any Helium, just curious from tech side what this new AWS service could offer. Not sure the downvotes are particularly about Helium or any crypto related discussion.

Why do peole care about helium so much?
People care about the value of their magic tokens and are spamming them in every online discussion in an effort to bring in more bag holders.
Agree. "awareness" drives people to own their token which will make them money.
Or because it solves an incentive problem which has kept millions of people with sub par networks to choose from.
Probably not if the Helium 5G offering is similar to their LoRaWan stuff.

Due to some security issues with the Helium LoRaWAN proof-of-coverage, only authorized hotspots/access points are allowed to participate. Helium manages this by issuing keys or certificates (IDK which it is) via the Decentralized Wireless Alliance[0].

I think it's highly unlikely Amazon would deploy Helium enabled 5G radios.

[0] https://dewi.org/

(comment deleted)
I'd love to see _some_ pricing estimates. There's a good amount I'd pay to get something like this for some rural communities, but it's unclear if I'm anywhere near able to afford it.
I would love to connect on this. Do you've any rural communities in mind ?
Probably a small outpost deployment + RF gear, so (guessing) $500K up front and $10-20k/mo. I don't know if it's possible to get RDOF grants for individual communities but that might cover some of it.
Aren't small cells significantly cheaper?
In addition to the gear, would you have to license the spectrum from someone? I think the fcc already auctioned it all off? Or does that not apply here? Any idea what that would cost?
I don’t know the details there. Presumably there would be at least be cost for someone to maintain the paperwork and equipment certification. There’s a couple comments in here about this being on CBRS, possibly more info there.
best thing is that you dont own the antenna, so Amz will prob rent out the remaining capacity.
That's not the way AWS typically prices things--my guess is the hardware and setup costs will be baked into the monthly service charges.
Generally yes but outpost is a little different:

https://aws.amazon.com/outposts/rack/pricing/

Either way this is just a guess :)

Even with Outposts you're not "buying" the rack (Amazon still owns it and takes it away at the end of the term), you're just committing to keeping it long enough for them to make money on it. The choice of paying upfront or paying monthly is roughly the same money management strategy available with any AWS service (e.g. EC2 Reserved Instances), and likely is more about lining up with capex/opex decisions than actual cost savings.
Totally agree 100%. Text communication sucks sometimes, if we were talking over beers it would have taken 3-5 seconds to clear up.
I think it'll be cheaper. We're looking to get the home unit cost to $499 & zero fees
What's next? Own your own island as a service™? I really like the direction AWS is headed and I like how they're opening up access to hard to setup hardware.
I have some experience building rural cellular networks (4G and starting to work on 5G), and can say that the core network part of the network is usually not the biggest challenge. There are open source cores available that actually work quite well for basic Internet access (magma has been mentioned elsewhere in the thread), and open5gs is another one. They can be deployed on lightweight edge infrastructure or in the cloud, since the computational overhead for the core is not huge for a small network (10s-100s of devices). SIM cards can be purchased pretty easily online from a variety of sources. There are even already existing turnkey solutions with a core network hosted in the cloud providing a management portal that integrates directly with like-branded radios (see Baicells).

Getting outdoor radios for rural access installed though is a bit more challenging, and I would be surprised if AWS was offering an outdoor solution here in the short term. Directional antennas and radio planning become a lot more important. There are a couple different players who will sell outdoor CBRS radios in small volume who all have pluses and minuses. CBRS is great for rural areas since there are often GAA channels available, but depending on the terrain may or may not provide huge area coverage. CBRS limits the height above average terrain and power you can deploy at. There are limits to the types of equipment and locations you can deploy without getting a professional installer certification. Getting the certification slightly raises those limits, but they are still something you need to take into account for wide-area access. You can actually get the CPI certification pretty easily via online classes offered by the different SAS (spectrum access service) providers. If you’re seriously considering founding a cellular wisp, there are some Facebook (unfortunately haha) groups out there that are pretty active and where you can get shoptalk questions answered about specific radios and technologies!

IMO the main value add from the AWS solution here is the access control, monitoring/auditing, and QoS management they are offering, which would be essential in an industrial setting, especially if running sensitive services over the network.

Just dropped you an email.
I wonder what the range is for their small cell towers. Would be interesting for off-grid/remote communities (pair with Starlink?).

edit: Reading more, it might not be the right use case - doesn't sound like something you use for your phone.

It's usable for phone but the biggest use case is data - Do you have any communities in mind
You can put the SIM in your phone but then you'll only be part of the private network. I didn't see anything about phone calls/SMS.
I wonder how many CISOs and CTOs trust Amazon.

I'm not sufficiently technical on this front and so I'm probably being a little on the side of paranoia but I don't trust Amazon with my cellular infrastructure.

What makes you trust telcos more?
Who said I trust telcos more?

Waveform.com is worth looking at.

I'd guess mostly their (telcos) incompetence.
They tend to have a track record, where this is a new offering and anyone signing up is ultimately the guinea pig?
CTOs of like half the companies on the planet (and 90% of the Fortune 500) which use AWS seem to trust it just fine.
How does this compare to Helium 5G? https://www.helium.com/5G
> Blockchain

Hard pass

It's actually not that bad. It's basically just a payment for providing verifiable wireless coverage to an area. If you want access to the network you buy it with VerizonBucks and people who provide coverage get them.

In this case the only reason for the blockchain is really because global payments are hard.

The biggest difference seems to be that aside from dropping some buzzwords like blockchain that page gives no description for what Helium 5G actually is or what the intended use cases for it are.
Kind of interesting how it stops being a buzzword when you understand why the word is being used, more interesting to watch other people be allergic to some words.

That landing page is sparse but Helium and other wireless networks are using blockchains as a rollout strategy.

Basically people invest in radio hardware because they think they can earn more in the blockchain token, don't worry as the radios are low power and most networks especially Helium's doesn't allow people to hoard radios, you earn less if there are other radios in the coverage zone. Other radios are the nodes which report your adherence to the network rules, the blockchain automatically pays out. Thats the supply side. The demand is people and organizations buying the token to buy send data over this network. With Helium I believe thats a one way conversion: Helium tokens -> data. But the receiving radio gets paid in new Helium from the blockchain's issuance, if it received any data then thats a bonus added to its payout allowance (but its not 1:1 to the amount of Helium tokens burned). So its kind of fun to think that the availability of this wireless network will have some service buying this digital commodity as an overhead cost and extending service to people that don't know Helium is one of the network routes behind the scenes. There are some IOT device that use the Helium network, I think some of the rideshare scooters use it already for over a year.

Helium 5G will become a large neutral host carrier for mobile using CBRS bands. Meaning that major network operators, or MVNOs, are the real customer.

The way this works is Helium hardware owner/operators deploy CBRS radios in whatever real estate assets they have at their disposal, then Helium or the Helium OEM partners with MNO/MVNO to offload data at a very cheap rate, and suddenly the carrier can grow network coverage at zero capex and low opex and the Helium node owner has a new source of revenue which was never before possible.

They also open the door to become a semi-private network infrastructure too, which would be in competition with this new Amazon product.

> wireless network supported by the Helium Blockchain.

The fuck

I mean it's very obvious how it's different: helium is not private. It's also obvious that you just wanted to namedrop Helium even though it has nothing to do with the use case for Amazon's product.
I beg to differ: it's a competitor for custom devices needing to transfer small amounts of data without regulatory hurdles, with the drawback of spotty connectivity given the current coverage map.

But for a moving device (ex: truck) transmitting locally stored data without latency concerns (ex: IOT temperature readings for a frozen cargo) I think it would be much more efficient, with far more coverage: you could have a $20 LORAWAN ESP32 built in individual boxes.

Think of this like Apple airtags, using a mesh network to intermittently transfer data on a best effort basis, but hooked to custom sensors so you fully control the payload

Helium is (initially) focused on shipping a LoRaWAN[0] network with a crowd sourced model for Lora RF hotspots that uses blockchain for to pay incentives to those supporting the network (via hardware, power, bandwidth). LoRaWAN focuses on low power, very low bandwidth, long range devices like sensor networks and simple controls.

Nearest competitor to Helium's LoRaWAN deployment that I'm aware of is The Things Network[1] which has no incentive model and instead people often host an access point for their personal use and sometimes for the use of those nearby.

Helium is expanding to CBRS[2] 5G service which will offer traditional higher bandwidth services, but this deployment is very early and the first units are shipping next month from FreedomFi[3].

Coolest thing about the Helium LoRaWAN offering is that you can buy a $20 ESP32 microcontroller with Lora radio on Amazon, write some software, buy Helium data credits and use it anywhere helium has coverage. No contracts, no special hardware. And if it sucks you could switch your application to use another LoRaWAN offering (The Things Network). That said, I think the coverage of Helium vastly outpaces anything else because of the crypto incentives fueling the madness and growth. [4]

[0] https://en.wikipedia.org/wiki/LoRa

[1] https://www.thethingsnetwork.org/

[2] https://en.wikipedia.org/wiki/Citizens_Broadband_Radio_Servi...

[3] https://freedomfi.com/

[4] https://explorer.helium.com/coverage

> Coolest thing about the Helium LoRaWAN offering is that you can buy a $20 ESP32 microcontroller with Lora radio on Amazon, write some software, buy Helium data credits and use it anywhere helium has coverage.

Or just go with the free TTN. But I do agree that it's good to have icentives and I'm quite glad somebody is rolling out a large scale LoRaWAN.

Helium has better coverage than TTN by orders of magnitude. I have an environmental sensor which reports every 15 minutes and $1 pays for years of data transfer.
It is similar concept but for AWS, you need to own your spectrum ($$$$).

Helium is a useless concept (in practice) that is now getting on the 5G trains for marketing purposes. Their IOT is super-low data rate for anything useful. It is good for people to thinker and speculate but no use for real life applications that demand actual throughput.

I wonder how do they link back to the main network? Would it require a fiber backbone in place as a prerequisite? Or through satellite uplink? i.e., would somewhere really remote, say, the oil rig in the North Sea, be out of luck for such installation?
It works over a regular IPSec tunnel I believe.
They could've called it AWS Infinidash Mobile.
There has never been any humor in Amazon's culture.
Probably worth noting that the regular cell providers can also provide SIMS that dump the end device into your existing, normal private network rather than the internet.

I don't know how this service differs in pricing, so it's hard to quantify when this AWS service would be a better idea outside of coverage issues.

I would assume that'd be so nickle and dimed by CISPs (probably trying to milk the enterprise cow) that nobody either knows or cares about that.
They do have to compete with one another, and potential customers usually already have leverage with other things they do with the same carrier, like WAN links, MPLS, corporate owned mobile contracts, etc.
Dedicated base stations are the difference here. There's a huge need for private communication networks outside of urban areas where there is bad cell coverage – think oil fields, docks, industrial estates, ranches, war zones.
Yes, sure...that's what I meant by "coverage issues". I wasn't sure everyone was aware of the route-to-my-private-network option that doesn't require a VPN, etc.
Is this a competitor to Twilio's managed SIM card service? That's been my go-to in recent years whenever we need to provide cell data connectivity to a small number of devices.
(comment deleted)
They love 5G because their mesh neighborhood spyveillance BS works even better without the need for WiFi permissions.
How long until Amazon takes on the existing cell companies?
Doesn't have the margins to make it attractive. An M2M/IoT MVNO though? That'd be cool to see, more competition in the space drives down prices.
Yea because amazon isn't used to entering markets with small margins.

/s

If mean, if AWS is dragging retail fulfillment around, do they need more anchors around their neck? Operating a nationwide cellular carrier is HARD, even when dumping the capex on tower management REITs and outsourcing all of your cellular infra to Ericsson. Seems easier to write a check and plug into someone else doing the schlep (ATT is best suited for this imho), maybe buy Twilio and have them do it.
Retail is basically cost-neutral at this point. I have a theory that they operate at a loss just to avoid taxes (because that delta is basically free) and spend that to stay extra competitive.

Amazon's whole MO is huge CapEx and selling it for cash flow. If they have the credit, why not finance it. They're already building satellite ground stations and a fleet to beam internet down to earth, this could easily supplement or build a market for that.

They'd probably benefit from lots of cheap and available internet for their logistics network, but its probably not expensive enough to justify alone, but maybe by leveraging AWS sales they can prop up cash flow enough to cover costs.

Not that they're doing this, of course, just that i don't think it should be ruled out.