malvcr
Posts: 50
Joined: Mon Apr 13, 2015 3:31 am

Re: Commercial Use of the Pi

Tue Mar 14, 2017 3:26 am

Well, the Zero wasn't really designed as you say, it was designed, really, to see how cheap it could be made. In fact the Zero isn't that great for IoT applications for one big reason - power consumption is too high. It cannot be battery powered for any length of time so needs to be hard wired to a power source, which makes for a bad IoT device for many use cases.

I agree that the "design goal" was "as cheap as possible", I do use a Pi Zero in a battery powered arrangement. To wit, an electronic name badge. A 4Ah "powerbank" will power a Pi Zero with a 3.5" display for about 14 hours on one charge. So...maybe not "IoT", but "wearable" anyway.
I rather suspect that it was a case of making the Pi Zero as cheap as possible, then throwing it out there to see what people would do with it. The headache is that a "$5 computer" caught peoples imaginations and the rest has been continuing rounds of trying--and often failing--to keep the pipeline from factory to retailer going as smoothly as possible. The low production rate has continued to fuel the supply problem, since anyone that thinks they have future use for multiple Pi Zeros is going to keep buying them whenever possible against that future need. Manufacturing *may* catch up when enough people realize that they have a decent sized box *full* of unused Pi Zeros on hand still waiting for that future use.
You see, I have been a software developer for several decades, and the last one I have been focused on security related issues (I am a Costa Rica founding member for the Association of cyber-security specialists). Then, I know how to use many different levels of computers (from very big to very small ones), and there is software I made that have been in charge of millions of financial transactions in some Banks. So I know what high-level enterprises need to work and when these needs are "exaggerated" or not well managed.

Then, I found the Raspberry machines. In fact, I was looking for something like this for a lot of years (from XX century) ... but no machine had the required specifications and price ratio until the Raspberry 2 appeared in the horizon. Now, I have several Raspberry Pi 2, 3, Zero, a Parallella 16, and a lot of Orange Pi computers in my personal laboratory performing many different tasks. For me, these machines have important purposes and when one machine stop being effective in one, then I put it in a different task. I have a big quantity of microSD cards and many versions of different operating system images.

Several months ago I found a "real-case" scenario for the Zero computer, and recently I had another one, very related with security tasks. When we talk about IoT as a concept, for me it is beyond current possibilities. This is why an Arduino it is very primitive from my perspective (my first attempt was with a Parallax machine years before the Arduino was invented). For some tasks a real computer it is needed ... but a PC it is not possible because of cost, size, heat, consumption. ARM based machines, in particular the Zero type are wonderful options. What happen is that IoT it is a very big umbrella that covers a wearable device can survive with direct solar energy and bigger computing devices that can be connected to the electricity grid. And the Zero is not the more consuming IoT device around.

In fact, my other alternative, being the Orange Pi Zero, needs some surgery and having a four core processor, consumes much more electricity than the Raspberry Pi Zero. And it is OK for the task (although the Raspberry has a better form factor).

If ... for me ... the Orange Pi Zero it is OK, it is important to observe that, after shipping, that machine cost more than $12 each unit. And after the courier cost and taxes from the US to Costa Rica, the computer go beyond $20 each unit .. and it still justify the investment. This is what I wanted to say in a previous post ... when people really need the Zero in enough quantities, the final price it is a secondary issue. But if the price go to second place, it is much better for the Foundation to have the money instead of making the shipping companies more rich, that it is what happens today. If I really need the Zero machines, I will make many tiny purchases (take into consideration that when adding international logistics this is a nightmare) ... so, at the end, the "one per customer" restriction it is useless.

And ... the CM it is not suited for this task and for many other tasks, because it needs electronic expertise not everybody have at hand. The Zero form factor it is just perfect.

Well ... a lot of words. I understand the origin and the current situation for the Zero; your explanation have been very clear and important for the availability discussion. I don't know if the Zero W will fix this, but today it doesn't seem to be the case ... no Zero, no Zero W are available, and both keep the one per customer restriction ... but let's see what happens :-)


Have a very good day

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 26370
Joined: Sat Jul 30, 2011 7:41 pm

Re: Commercial Use of the Pi

Tue Mar 14, 2017 9:11 am

ZeroW has only been out a couple of weeks so demand is till high. I expect, but I have no particular knowledge, that it will be available in larger quantities simply because of the better profit margins. So that should mean, once demand is sated, that the one per customer restriction will be lift. When? Don't know.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed.
I've been saying "Mucho" to my Spanish friend a lot more lately. It means a lot to him.

fruitoftheloom
Posts: 23021
Joined: Tue Mar 25, 2014 12:40 pm
Location: Delightful Dorset

Re: Commercial Use of the Pi

Tue Mar 14, 2017 9:23 am

jamesh wrote:ZeroW has only been out a couple of weeks so demand is till high. I expect, but I have no particular knowledge, that it will be available in larger quantities simply because of the better profit margins. So that should mean, once demand is sated, that the one per customer restriction will be lift. When? Don't know.
The essence of this post has been lost.

Yes when available the Raspberry Pi Zero W can be used in a Commercial Product but it must be noted that long term availability is not guaranteed, the Zero W is the 3rd model of the Zero family in 18 months.

Therefore the Compute Module and a Custom Carrier Board is the way forward if you still want to offer a product 5 years down the line.......
Rather than negativity think outside the box !
RPi 4B 4GB (SSD Boot)..
Asus ChromeBox 3 Celeron is my other computer...

RareHare
Posts: 89
Joined: Thu Jun 20, 2013 7:17 pm

Re: Commercial Use of the Pi

Sat Mar 18, 2017 11:53 pm

Having read all of this thread, especially t\he comments where individuals keep saying that it is not economical to scale-up production for the Pi Zero W, one is left scratching his head. In general, in the world of PCB manufacturing, just the opposite is true: The more you produce, the lower is the cost. This generalization applies even in many cases where new production equipment has to be added to accommodate increased capacity. For example: One might add a $500,000 investment for new pick-and-place/reflow/rework equipment, with the idea that the expense will be amortized after a certain number of units have shipped.

It is my personal opinion that those in-the-know, regarding the Pi Zero W, know good-and-well that there is significant demand for the Pi Zero W, at $10, AS IS, with all its warts and flaws. Saying that It was never meant for... is like telling a jeweler...You really don't want these raw diamonds...they are really, really dirty...it is hard to see the sparkle. The jeweler knows what he is getting, and is able to determine for himself whether the dirty diamonds are appropriate for the cost/application combination.

With that said, one cannot help but wonder if the real problem is not {whatever is being stated as the problem}, but that the margin is actually (slightly) negative. I have not done a bill-of-materials (BOM) on the Pi Zero, nor the Pi Zero W, but someone should. It would be instructive to do the BOM, and get a cost for PCB+parts+assembly, in assembly quantities of 100,000. The goal would be to see if it is possible to get down to $5 for the Pi Zero and $10 for the PI Zero W.

IF, this is the actual problem, that the Pi Zero and Pi Zero W are actually loss-leaders, and the Foundation is hoping to recoup said losses by pushing the CM whenever someone ask about bulk purchases of the Pi Zero/W, I think the Foundation should simply say that, instead of saying that the problem is something else.
Last edited by RareHare on Sun Mar 19, 2017 12:52 am, edited 2 times in total.

fruitoftheloom
Posts: 23021
Joined: Tue Mar 25, 2014 12:40 pm
Location: Delightful Dorset

Re: Commercial Use of the Pi

Sun Mar 19, 2017 12:00 am

RareHare wrote:Having read all of this thread, especially t\he comments where individuals keep saying that it is not economical to scale-up production for the Pi Zero W, one is left scratching his head. In general, in the world of PCB manufacturing, just the opposite is true: The more you produce, the lower is the cost. This generalization applies even in many cases where new production equipment has to be added to accommodate increased capacity. For example: One might add a $500,000 investment for new pick-and-place/reflow/rework equipment, with the idea that the expense will be amortized after a certain number of units have shipped.

It is my personal opinion that those in-the-know, regarding the Pi Zero W, know good-and-well that there is significant demand for the Pi Zero W, at $10, AS IS, with all its warts and flaws. Saying that It was never meant for... is like telling a jeweler...You really don't want these raw diamonds...they are really, really dirty...it is hard to see the sparkle. The jeweler knows what he is getting, and is able to determine for himself whether the dirty diamonds are appropriate for the cost/application combination.

With that said, one cannot help but wonder if the real problem is not {whatever is being stated as the problem}, but that the margin is actually (slightly) negative. I have not done a bill-of-materials (BOM) on the Pi Zero, nor the Pi Zero W, but someone should. It would be instructive to do the BOM, and get a cost for PCB+parts+assembly, in assembly quantities of 100,000. The goal would be to see if it is possible to get down to $5 for the Pi Zero and $10 for the PI Zero W.

IF, this is the actual problem, that the Pi Zero and Pi Zero W are actually loss-leaders, and the Foundation is hoping to recoup said losses by pushing the CM whenever someone ask about bulk purchases of the Pi Zero/W, I think the Foundation should simply say that, instead of saying that the problem is something else.
You forgot to factor in that only a single Fab Plant manufactures the BCM2835 SoC which is now used in 5 models.
Rather than negativity think outside the box !
RPi 4B 4GB (SSD Boot)..
Asus ChromeBox 3 Celeron is my other computer...

RareHare
Posts: 89
Joined: Thu Jun 20, 2013 7:17 pm

Re: Commercial Use of the Pi

Sun Mar 19, 2017 12:50 am

fruitoftheloom wrote:You forgot to factor in that only a single Fab Plant manufactures the BCM2835 SoC which is now used in 5 models.
The same principle applies. Whenever there is an under-served market of, say, 5,000,000+ of an SoC, the owner of the intellectual property generally has no problem finding and extra fab to take on the extra work. My conservative estimate of the Pi Zero W, as it is, is 25,000,000 units.

User avatar
davidcoton
Posts: 4872
Joined: Mon Sep 01, 2014 2:37 pm
Location: Cambridge, UK
Contact: Website

Re: Commercial Use of the Pi

Sun Mar 19, 2017 12:58 am

RareHare wrote:Having read all of this thread, especially t\he comments where individuals keep saying that it is not economical to scale-up production for the Pi Zero W, one is left scratching his head.
I don't think you have read carefully enough. No-one has said it is not economical to scale up production, unless I missed it :o . Just that it is not commercially desirable to sell Pi Zeros more than one at a time. I think you may find that the economies of scale (beyond current production) are very small, and may be non-existent if Pi Zero production is actually filling gaps in the production schedule of other boards, so is being charged by Sony at cost or little more.

The Pi Zero is, if not a loss leader, a very low profit item. RPF/T have decided to limit production, for reasons they have not made public. But we can guess: even with increased production, the margin is too low. they designed it as an exercise in producing a low-cost SBC. RPF/T try very hard not to increase the dollar price of their products once launched, and to do so with the Pi Zero would conflict with the lowest-possible-cost aim. The success is almost embarrassing, and scaling up is not an option because of the effects on production and/or sales of other Pis. In any case RPF/T have no interest in selling at low/no profit just to boost the profits of other commercial enterprises. I don't know what RPT would say if you offered to buy 100,000 Pi Zeros at £10 each. That might change things around a bit. Note that one retailer (at least) tried multiple sales with effectively negative quantity discounts -- the more you buy, the higher the unit price. I don't know what conclusions were drawn from that attempt.

The Pi Zero W appears to have a significantly increased margin. We don't know, but it appears that the restriction to single-unit sales may be temporary. Meanwhile it has the side benefit of making life harder for scalpers. Multiple unit sales may become possible, and are in any case desirable for the education market. When production can sustain the demand, the limit can be raised (maybe not removed) without a secondary market bonanza.

When selecting a component for a product, price and specification are important. But in many cases guaranteed availability is also important. RPF/T do not have the profit margin on the Pi Zero (and probably not on the Pi Zero W) to make that commitment. It may not be important to you, you may want to make a single batch and not guarantee future availability or maintainability of your product. In that case one of these models suits you. However you may (I don't know your business as well as you do) walk into a minefield where your reputation is at stake if unexpected repeat orders arise. The advice to avoid using the Pi Zero (W) in commercial products is given in good faith by people who are aware of one side of this issue. YMMV.
My conservative estimate of the Pi Zero W, as it is, is 25,000,000 units.
Way out. Total Pi production (all models) is about half that to date. Maybe 25 million Pi Zero W over the next few years, but if it's that successful expect a revised version anyway.
Signature retired

RareHare
Posts: 89
Joined: Thu Jun 20, 2013 7:17 pm

Re: Commercial Use of the Pi

Sun Mar 19, 2017 2:05 am

davidcoton wrote: No-one has said it is not economical to scale up production, unless I missed it :o . Just that it is not commercially desirable to sell Pi Zeros more than one at a time.
That is the meaning of scaling-up. Either it is, or it is not. If it is economically-viable to sell one at a time, in most cases, it is more economically-viable to sell more at a time.
I think you may find that the economies of scale (beyond current production) are very small, and may be non-existent if Pi Zero production is actually filling gaps in the production schedule of other boards, so is being charged by Sony at cost or little more.
IMHO, The Foundation should state that, then. They should state: This product would be sold as a loss, were we not using idled machines at Sony. It is not a viable product in its own right at the current price, even if we were to sell a million units.
The Pi Zero is, if not a loss leader, a very low profit item. RPF/T have decided to limit production, for reasons they have not made public. But we can guess: even with increased production, the margin is too low. they designed it as an exercise in producing a low-cost SBC. RPF/T try very hard not to increase the dollar price of their products once launched, and to do so with the Pi Zero would conflict with the lowest-possible-cost aim. The success is almost embarrassing, and scaling up is not an option because of the effects on production and/or sales of other Pis.
The Foundation should say that then, IMO.
In any case RPF/T have no interest in selling at low/no profit just to boost the profits of other commercial enterprises.
No, they don't. But we do. :twisted: Actually, I am not saying that they should. I am saying that they could clear up a huge amount of head-scratching if they would be forthcoming for the true reason that they are not meeting obvious demand, for both personal and commercial.
I don't know what RPT would say if you offered to buy 100,000 Pi Zeros at £10 each. That might change things around a bit. Note that one retailer (at least) tried multiple sales with effectively negative quantity discounts -- the more you buy, the higher the unit price. I don't know what conclusions were drawn from that attempt.
Interesting.
The Pi Zero W appears to have a significantly increased margin. We don't know, but it appears that the restriction to single-unit sales may be temporary. Meanwhile it has the side benefit of making life harder for scalpers.
That is not hard to fix. If the law in a country allows "limit-one-per-customer", those same laws allow prioritization of consumer sales over reseller sales. A scalper will be far less eager to try to build a business around scalping product if they know that they will be undercut by their own supplier using a prioritization scheme toward individuals and non-resellers. The supplier would allow the individual customers to purchase on back-order. As soon as new orders arrived, priority would be given to individuals and resellers who sell at dictated cost ($10). In the USA, for example, a manufacturer is allowed to print the cost of a product directly onto the product to prevent reseller from gouging. It would not be hard to catch resellers attempting to resale Pi Zero W's for more than $10 naked.
When selecting a component for a product, price and specification are important. But in many cases guaranteed availability is also important. RPF/T do not have the profit margin on the Pi Zero (and probably not on the Pi Zero W) to make that commitment. It may not be important to you, you may want to make a single batch and not guarantee future availability or maintainability of your product. In that case one of these models suits you. However you may (I don't know your business as well as you do) walk into a minefield where your reputation is at stake if unexpected repeat orders arise. The advice to avoid using the Pi Zero (W) in commercial products is given in good faith by people who are aware of one side of this issue. YMMV.
I agree, but I have a feeling that the non-scalpers who have wanted to buy in bulk would be still be ready to buy, despite this disclaimer. Also, on this point, the nature of markets is such that, a profitable product, for significant demand, almost always meets the market. This is what lead me to believe that the true problem is negative-profitability. I realize that the Foundation's goal is not to enrich itself on our beloved RPi's, but if I were to try to explain the rationale given for not meeting the market to most people in business, (actually I spent probably 8 or 9 hours to my investors), they would tell me that something does not smell right.
My conservative estimate of the Pi Zero W, as it is, is 25,000,000 units.
Way out. Total Pi production (all models) is about half that to date. Maybe 25 million Pi Zero W over the next few years, but if it's that successful expect a revised version anyway.
Let's hope so. It would be the first time that a beefy, generally-available, generalized, computer were sold <= $10. One of the major barriers to mass-proliferation of IoT would have been shattered.

W. H. Heydt
Posts: 12319
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: Commercial Use of the Pi

Sun Mar 19, 2017 3:05 am

RareHare wrote:
fruitoftheloom wrote:You forgot to factor in that only a single Fab Plant manufactures the BCM2835 SoC which is now used in 5 models.
The same principle applies. Whenever there is an under-served market of, say, 5,000,000+ of an SoC, the owner of the intellectual property generally has no problem finding and extra fab to take on the extra work. My conservative estimate of the Pi Zero W, as it is, is 25,000,000 units.
Over how much time? There have been about 500K Pi Zeros sold in about 15 months, plus--perhaps--150K Pi Zero Ws in 3 weeks, but the last figure includes the 80K pre-launch boards.

Now *if* it took, say, 3 times the current production rates to keep the retail pipeline filled to the point that ordering multiple Pi Zeros and Pi Zero Ws could work, you'd be talking about 3*(12K+25K) per week ~ 450K combined production per month. At that rate, and note that it is rather higher than the sales rate (which is probably very close to the production rate) if the Pi3B, which sold 4 million units in its first year, it would take about 5 years to sell that 25M units. Note also, that those figures suggest an approximate *doubling* of the total annual production of Pis of all other varieties.

RareHare
Posts: 89
Joined: Thu Jun 20, 2013 7:17 pm

Re: Commercial Use of the Pi

Sun Mar 19, 2017 4:01 am

W. H. Heydt wrote:Over how much time?
I was thinking 5-7 years.
There have been about 500K Pi Zeros sold in about 15 months, plus--perhaps--150K Pi Zero Ws in 3 weeks, but the last figure includes the 80K pre-launch boards.

Now *if* it took, say, 3 times the current production rates to keep the retail pipeline filled to the point that ordering multiple Pi Zeros and Pi Zero Ws could work, you'd be talking about 3*(12K+25K) per week ~ 450K combined production per month. At that rate, and note that it is rather higher than the sales rate (which is probably very close to the production rate) if the Pi3B, which sold 4 million units in its first year, it would take about 5 years to sell that 25M units. Note also, that those figures suggest an approximate *doubling* of the total annual production of Pis of all other varieties.
Should the 12K be roughly 8K? That's what I get from 500K/15/4, if that is what you mean. In any case, 4 million for Pi3B is outstanding. It is a bit shocking. :o I think such success says all.

We have to remember:

There is a enormous amount of pent-up demand for a Pi Zero W - like computer.

This is not a situation where a novel product appears that people gradually find interesting. This is a situation where a large number of people have been sitting, waiting, preying for the right computer to come along. This demand goes back to the early days of the Internet. Cisco Systems has been talking about the Internet of Things forever. Even when the Intel Edison came out, there was a lot of excitement, which quickly turned into frustration, because Intel kept playing games with price/availability and not being straightforward with would-be customers. One gets the feeling that they made their announcement prematurely to steal thunder from the Raspberry Pi.

Aside from the disappointment of availability, there have been three major barriers to committing to an IoT hardware platform:
  1. The cost was too high for a device that provides all of the goodies (powerful CPU, RAM+ROM in the megabytes, wireless link, generous bank of interfaces [GPIO, etc.]).
  2. The power-consumption was too great for the device.
  3. The protocol stack, TCP/IP, was the equivalent of propeller-driven airplane in the age of jet-engines.
With the Pi Zero W:
  1. The first problem is eliminated.
  2. It is likely that the power problem is solvable on a case-by-case basis.
  3. The last problem is being solved as I type this by researchers in the field.
The only other device on the market that offers significant competition ($9 computer does not count, IMHO); is the ESP32 by Espressif. But the ESP32, despite its great power-management features, cannot compete with the Pi Zero W on performance. The RAM alone in the Pi Zero W destroys this ~$5.00 computer.

Therefore, one can conclude that, right now, the Pi Zero W is the computer for IoT, or at least it should be.

W. H. Heydt
Posts: 12319
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: Commercial Use of the Pi

Sun Mar 19, 2017 4:46 am

RareHare wrote:
W. H. Heydt wrote:Over how much time?
I was thinking 5-7 years.
Note that at the 5 year time scale, more Pi Zero/Pi Zero W boards would have to be made per year than total Pi production per year now. The 25M is twice the number of Pis--of *all* types--than have been sold to date. I'm not saying that the market can't absorb them, necessarily, but the numbers represent a monummental increase in manufacturing capacity.
There have been about 500K Pi Zeros sold in about 15 months, plus--perhaps--150K Pi Zero Ws in 3 weeks, but the last figure includes the 80K pre-launch boards.

Now *if* it took, say, 3 times the current production rates to keep the retail pipeline filled to the point that ordering multiple Pi Zeros and Pi Zero Ws could work, you'd be talking about 3*(12K+25K) per week ~ 450K combined production per month. At that rate, and note that it is rather higher than the sales rate (which is probably very close to the production rate) if the Pi3B, which sold 4 million units in its first year, it would take about 5 years to sell that 25M units. Note also, that those figures suggest an approximate *doubling* of the total annual production of Pis of all other varieties.
Should the 12K be roughly 8K? That's what I get from 500K/15/4, if that is what you mean. In any case, 4 million for Pi3B is outstanding. It is a bit shocking. :o I think such success says all.
No. If the 12K is off, it should be 12.5K. The first few months of Pi Zero production were fairly low, and from around mid-January 2016 until into March 2016, none were made at all because of the ramp up, launch, and initial surge of Pi3B production. After that, Dr. Upton stated a goal of making 50K per month. I am assuming that that rate has been achieved. And while my numbers project that forward, I am not at all sure that anywhere near that many are being produced right now, as evidenced by the number of vendors who are out of Pi Zeros and don't appear to be getting restocked.

FYI...I don't find the 4 million Pi3Bs in the first year at all surprising. Last Fall when it was announced that 10 million Pis had been sold, someone in the blog comments asked how long it had taken to sell the most recent million. The answer was: 3 months. So even then, the overall rate of Pis sales was running at 4 million per year.
We have to remember:

There is a enormous amount of pent-up demand for a Pi Zero W - like computer.
Maybe... I have yet to see anything that could actually justify that large a market. It's a nice device and I have one application (already implemented, but I may make a few units) and can do tour de force "look how small I can make this useful thing" (something I've also done, and the one unit is all I ever expect to have a use for), that's a very limited market. Other than that, I suspect the biggest real use is likely to be DIY robots, which can be a decent market, but not on the scale you're expecting.
This is not a situation where a novel product appears that people gradually find interesting. This is a situation where a large number of people have been sitting, waiting, preying for the right computer to come along. This demand goes back to the early days of the Internet. Cisco Systems has been talking about the Internet of Things forever. Even when the Intel Edison came out, there was a lot of excitement, which quickly turned into frustration, because Intel kept playing games with price/availability and not being straightforward with would-be customers. One gets the feeling that they made their announcement prematurely to steal thunder from the Raspberry Pi.
I am less than convinced that "IoT" is or is going to be all that much more than marketing hype. For some of the touted uses of IoT (e.g. remote control lights) a Pi Zero is far too big.
Aside from the disappointment of availability, there have been three major barriers to committing to an IoT hardware platform:
  1. The cost was too high for a device that provides all of the goodies (powerful CPU, RAM+ROM in the megabytes, wireless link, generous bank of interfaces [GPIO, etc.]).
  2. The power-consumption was too great for the device.
  3. The protocol stack, TCP/IP, was the equivalent of propeller-driven airplane in the age of jet-engines.
With the Pi Zero W:
  1. The first problem is eliminated.
  2. It is likely that the power problem is solvable on a case-by-case basis.
  3. The last problem is being solved as I type this by researchers in the field.
The only other device on the market that offers significant competition ($9 computer does not count, IMHO); is the ESP32 by Espressif. But the ESP32, despite its great power-management features, cannot compete with the Pi Zero W on performance. The RAM alone in the Pi Zero W destroys this ~$5.00 computer.

Therefore, one can conclude that, right now, the Pi Zero W is the computer for IoT, or at least it should be.
I suspect that the devotees of other boards could as easily dismiss the Pi Zero W as you are dismissing those other boards. *If* IoT sticks around and comes to pass as you expect, I rather suspect that the Pi Zero W would be seen as bloated, power hungry, and utter overkill as regards processor power and memory. Why would an IoT device need to run a full Linux OS, for instance? Wouldn't bare metal programming be better, require far less RAM, consume less power, and just need to report data back to a larger system for analysis?

RareHare
Posts: 89
Joined: Thu Jun 20, 2013 7:17 pm

Re: Commercial Use of the Pi

Sun Mar 19, 2017 4:59 am

W. H. Heydt wrote:I suspect that the devotees of other boards could as easily dismiss the Pi Zero W as you are dismissing those other boards. *If* IoT sticks around and comes to pass as you expect, I rather suspect that the Pi Zero W would be seen as bloated, power hungry, and utter overkill as regards processor power and memory. Why would an IoT device need to run a full Linux OS, for instance? Wouldn't bare metal programming be better, require far less RAM, consume less power, and just need to report data back to a larger system for analysis?
In a nutshell: It's the protocol stack. Generalized IoT requires several features, the most notable of which are:
  1. Generalized mobility: As node moves down street, connections are made and broken with WiFi access points distributed along the street.
  2. Generalized symmetric/asymmetric distributed cryptosystem so that connections between any two applications on the Internet gets security for free without much forethought by programmer.
  3. Generalized multicast - If your device is being used as consumer of streamed information, the global stream should be optimal.
  4. Generalized Routing - Routing just works, and routing tables configure themselves dynamically and optimally.
  5. etc.
Perhaps you will take me at my word if I told you that, to sold the major software problems generalized computer networking, which is the reason that IoT has not reached its potential, IMO, one needs about 800KB of program memory for the protocol stack alone. The CPU must support true hardware context switching and multi-threading (if it is to be done right). A hardware based random-number generator helps, and RAM needs to be significant: 64KB is not enough if one hopes to do a simple recompile without tailoring toward the hardware.

So yes, Pi Zero W is both overkill, but necessary to do it "right". If one wants to do one-off, tweak here, tweak there, chop here, pack there; applications, then one might be better using ESP32 or something like that.

But one should not underestimate the power of a WiFi-enabled, fully generalized computer, with copious amounts of ROM and RAM being sold at $10. I wish I could give some actual examples of utterly-fantastic applications, but I am under N.D.A.

User avatar
davidcoton
Posts: 4872
Joined: Mon Sep 01, 2014 2:37 pm
Location: Cambridge, UK
Contact: Website

Re: Commercial Use of the Pi

Sun Mar 19, 2017 9:22 am

RareHare wrote:IMHO, The Foundation should state that, then. They should state: This product would be sold as a loss, were we not using idled machines at Sony. It is not a viable product in its own right at the current price, even if we were to sell a million units.
...
The Foundation should say that then, IMO.
...
I am saying that they could clear up a huge amount of head-scratching if they would be forthcoming for the true reason that they are not meeting obvious demand, for both personal and commercial.
RareHare wrote:I wish I could give some actual examples of utterly-fantastic applications, but I am under N.D.A.
So, you think that RPF should disclose their commercial strategy, while you hide behind an NDA?
Just accept that you are not going to be able to buy multiple Pi Zeros at $5, ever. You are not yet able to buy multiple Pi Zero Ws, that may happen, possibly even at $10, but it is not guaranteed. Nor is future supply.

And consider again the power requirements. Pi Zeros as of now are far too power hungry for battery-based IOT. If the device is wired for power, the cost of installation dwarfs the hardware cost. And if you implement the hardware to power down the Pi, again it will cost more than the Pi itself.

Finally, if you think you can build a $5 or even $10 SBC with WiFi, good IoT software and hardware support, reasonably powerful yet low in power consumption, and selling 25 million over five years, then you are wasting your time here. Go do it, make your fortune and be famous. You won't be competing with RPF because that is not their market.
Signature retired

hippy
Posts: 7346
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: Commercial Use of the Pi

Sun Mar 19, 2017 3:08 pm

RareHare wrote:There is a enormous amount of pent-up demand for a Pi Zero W - like computer.
Absolutely agree; just as there was a enormous amount of demand for the original Pi. From the very beginning I have said the Foundation massively underestimates Pi potential and demand.
RareHare wrote:Therefore, one can conclude that, right now, the Pi Zero W is the computer for IoT, or at least it should be.
I would say the Zero W is the best option available for IoT at present. But it's not perfect and probably will not ultimately be the product of choice for IoT commercially.

Its four main problems are; its high power consumption, its lack of battery and low-power management, the potential for SD Card corruption on loss of power, and it has not been designed or intended as a product for commercial use.

It is a great product for stepping out into the IoT arena with, for experimenting and prototyping IoT and the related infrastructure, but it is not well suited to commercial IoT ventures.

And please note that none of that is criticism of what the Zero W actually is.

PiGraham
Posts: 3926
Joined: Fri Jun 07, 2013 12:37 pm
Location: Waterlooville

Re: Commercial Use of the Pi

Sun Mar 19, 2017 3:50 pm

hippy wrote:
RareHare wrote:There is a enormous amount of pent-up demand for a Pi Zero W - like computer.
Absolutely agree; just as there was a enormous amount of demand for the original Pi. From the very beginning I have said the Foundation massively underestimates Pi potential and demand.
RareHare wrote:Therefore, one can conclude that, right now, the Pi Zero W is the computer for IoT, or at least it should be.
I would say the Zero W is the best option available for IoT at present. But it's not perfect and probably will not ultimately be the product of choice for IoT commercially.

Its four main problems are; its high power consumption, its lack of battery and low-power management, the potential for SD Card corruption on loss of power, and it has not been designed or intended as a product for commercial use.

It is a great product for stepping out into the IoT arena with, for experimenting and prototyping IoT and the related infrastructure, but it is not well suited to commercial IoT ventures.

And please note that none of that is criticism of what the Zero W actually is.
Since you identified four area of weakness, and other products exist that beat it in all four, PiZW is not " the best option available for IoT at present". It may be the best for some niches in IoT, where you need a hi-res display or a full Linux, but there are surely lots of IoT projects that don't need that, where deterministic performance is important or battery power is critical, where something else is a better choice.

I love the PiZ and PiZW. I have two of each and I can see a lot of potential, but it doesn't suit everything IoT.

RareHare
Posts: 89
Joined: Thu Jun 20, 2013 7:17 pm

Re: Commercial Use of the Pi

Sun Mar 19, 2017 6:29 pm

davidcoton wrote:So, you think that RPF should disclose their commercial strategy, while you hide behind an NDA?
That's just it. I did not know that RPF was a "commercial entity", in the traditional sense. I have no problem one way or the other. My company is not in competition with RPF in any way or other, and never would be. My relationship with RPF could be summarized as:

Hey, I was in your store trying to find your XYZ widgets, and I noticed that you are out...I need 25 of them. Do you know when I will be able to get 25?
Just accept that you are not going to be able to buy multiple Pi Zeros at $5, ever. You are not yet able to buy multiple Pi Zero Ws, that may happen, possibly even at $10, but it is not guaranteed. Nor is future supply.
Says who? And also, again, I have no problem with that. Actually, I have no problems with RPF at all. I am merely saying that the reason that was given does not compute.
And consider again the power requirements. Pi Zeros as of now are far too power hungry for battery-based IOT. If the device is wired for power, the cost of installation dwarfs the hardware cost. And if you implement the hardware to power down the Pi, again it will cost more than the Pi itself.
I invoke my Jeweler's Analogy as stated a few posts backward. Telling someone:

We are not going to sell you this because it is probably something that you do not want.

...does not make any sense. It is as if the entity selling the thing knows better than the buyer what is good for the buyer. The determination of what is good and bad for the buyer, in this particular case, easily rests with the buyer.
Finally, if you think you can build a $5 or even $10 SBC with WiFi, good IoT software and hardware support, reasonably powerful yet low in power consumption, and selling 25 million over five years, then you are wasting your time here. Go do it, make your fortune and be famous. You won't be competing with RPF because that is not their market.
I never said I could. Nor did I say anyone could. On the contrary, I said that it might be exceptionally difficult to do, if not impossible.The only thing I have said so far is that the reason for not being able to scale-up, as given, does not make sense, given the market for the Pi Zero W.

hippy
Posts: 7346
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: Commercial Use of the Pi

Sun Mar 19, 2017 7:04 pm

PiGraham wrote:Since you identified four area of weakness, and other products exist that beat it in all four, PiZW is not " the best option available for IoT at present".
Perhaps I should have clarified "I would say the Zero W is the best option available for IoT at present" with "given the weakneses in the alternatives; price and community support in particular".

People will have to weigh-up all aspects of the alternatives, what they need, want and would like, to determine what is best for them.
Last edited by hippy on Sun Mar 19, 2017 7:05 pm, edited 1 time in total.

RareHare
Posts: 89
Joined: Thu Jun 20, 2013 7:17 pm

Re: Commercial Use of the Pi

Sun Mar 19, 2017 7:05 pm

I was just thinking that it is conceivable that RPF knew all along that it would never allow Pi Zero to sell in quantity, regardless of demand. The strategy would work something like this:
  1. Make an announcement of Pi Zero being sold at $5.
  2. Accept that there will be a loss for each unit sold.
  3. Restrict sales to one or a few per customer, because cumulative losses would have to be kept under a threshold.
  4. Let the technical media latch-on. They will print: A Full-Blown Computer That Runs Linux Is Now Available For $5
  5. When people (corporations) try to buy in quantity, steer them toward the higher-priced Compute Module.
  6. Hope that the rate-of-profit of the Compute Module compensates for the rate-of-loss of the Pi Zero.
  7. Add WiFi to Pi Zero, but do not let price of Pi Zero W merely reflect increase in BOM for WiFi chip. Add a bit more: Double price of entire unit to $10.
This strategy would allow the RPF to achieve its objective in supporting students, hobbyists, universities, etc. in getting at least one per individual. Those corporations that decide to munch on the Compute Module would be essentially subsidizing the sells of $5 and $10 computers to the not-as-privileged.

Two things to note:
  1. This supposition is 100% pure speculation.
  2. Were it true, I would certainly have no problem with it, and I expect that very few other Friends-of-the-Foundation would.
To reiterate, if RPF came-out and said...

Hey guys, truth be told, we eat $X for every Pi Zero and Pi Zero W sold, and the reason that retailers are not allowing you to buy from them in quantity, is because we will not, and have no intention to, allow them to buy from us in quantity sufficient to meet demand. The units have negative profit margin, and we are counting on sells of the Compute Module, especially to commercial entities, to make-up for the losses.

...I would have no problem with that.
Last edited by RareHare on Sun Mar 19, 2017 7:57 pm, edited 3 times in total.

hippy
Posts: 7346
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: Commercial Use of the Pi

Sun Mar 19, 2017 7:08 pm

RareHare wrote:Accept that there will be a loss for each unit sold.
There is no evidence that each Zero makes a loss. It doesn't generate enough profit for it to be made in huge quantity which is very different to making a loss. Even so they have manufactured some 600K units.

Heater
Posts: 15777
Joined: Tue Jul 17, 2012 3:02 pm

Re: Commercial Use of the Pi

Sun Mar 19, 2017 7:17 pm

RareHare,
...the reason for not being able to scale-up, as given, does not make sense,...
Who are you to know such a thing?

The Pi Foundation knows what resources are available to them and how to make best use of such resources. As a charitable foundation their mission is not to make profit or provide cheap computers to the world for other companies to profit from. They are an educational charity.

If you know better how to make such things for commercial use then we look forward to your alternative product. I'm sure there are many around the world working on this as we speak.
Memory in C++ is a leaky abstraction .

User avatar
rpdom
Posts: 16964
Joined: Sun May 06, 2012 5:17 am
Location: Chelmsford, Essex, UK

Re: Commercial Use of the Pi

Sun Mar 19, 2017 7:20 pm

RareHare wrote:The only thing I have said so far is that the reason for not being able to scale-up, as given, does not make sense, given the market for the Pi Zero W.
Actually, it does make sense.

As you know, the RPF is a charity. Their aim is to get people (of all ages, but primarily youngsters like me(?)) programming. They continuously try to push boundaries within their budget and the Pi Zero is more a proof of concept than a commercial device. It wasn't designed to be produced in huge quantities. Ramping up production wouldn't be viable. A new production line would be needed and the cost of that probably wouldn't be covered even by ten years of Pi Zero (W) production even if enough BCM2385 chips could be sourced (I suspect it wouldn't be worth setting up a new fab for the numbers required).

Also you mentioned a 7-year payback to cover the costs, well, it has been hinted at that the expected production life for any given Pi model will be around 3 years (apart from the CM versions which are aimed at the commercial market).

RareHare
Posts: 89
Joined: Thu Jun 20, 2013 7:17 pm

Re: Commercial Use of the Pi

Sun Mar 19, 2017 8:11 pm

The RPF could stop all the back-and-forth with a simple statement on the Pi Zero page:

We have had numerous individuals and corporations ask about purchasing the Pi Zero in quantity. We will never sell the Pi Zero in quantity. Thank You.

The technical media will spread the word:

Caveat: The Pi Zero will never be available for purchase in significant quantity.

Then, every person/entity who reads: $5 Linux Computer...blah blah blah... will also see this disclaimer because the technical media will print it. Expectations will be set, and thoughts about purchasing in quantity will be nipped in the bud.

As it stands, that is not what is happening. As it stands, people/corporations see... $5 Computer blah..blah..blah...... and presume, reasonably, that they can by 10 for $50.
Last edited by RareHare on Sun Mar 19, 2017 8:50 pm, edited 1 time in total.

Heater
Posts: 15777
Joined: Tue Jul 17, 2012 3:02 pm

Re: Commercial Use of the Pi

Sun Mar 19, 2017 8:30 pm

RareHare,
As it stands, people/corporations see... $5 Computer blah..blah..blah...... and presume, reasonably, that they can by 10 for $50.
Not in my decades of experience.

Corporations, commercial users, contact their supplier and ask for a quote on price and delivery of what ever quantities they need. For now, and into the future. It would be foolish to design a product around a component you cannot get for a good price and in the quantities you desire.

Meanwhile, "people" who are interested in such things know the situation with the Raspberry Pi.
Memory in C++ is a leaky abstraction .

RareHare
Posts: 89
Joined: Thu Jun 20, 2013 7:17 pm

Re: Commercial Use of the Pi

Sun Mar 19, 2017 8:43 pm

I thought you were going to say that. 8-) Of course no one designs a product around an unknown product. That is not the point. The point is that, any time spent, even if it is a few hours, that is time wasted unnecessarily. Spread this across tens of thousands of people/entities who follow the same path, and you get the idea.

W. H. Heydt
Posts: 12319
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: Commercial Use of the Pi

Sun Mar 19, 2017 9:01 pm

RareHare wrote:
davidcoton wrote:So, you think that RPF should disclose their commercial strategy, while you hide behind an NDA?
That's just it. I did not know that RPF was a "commercial entity", in the traditional sense. I have no problem one way or the other. My company is not in competition with RPF in any way or other, and never would be. My relationship with RPF could be summarized as:

Hey, I was in your store trying to find your XYZ widgets, and I noticed that you are out...I need 25 of them. Do you know when I will be able to get 25?[quo[e]Just accept that you are not going to be able to buy multiple Pi Zeros at $5, ever. You are not yet able to buy multiple Pi Zero Ws, that may happen, possibly even at $10, but it is not guaranteed. Nor is future supply.

To the extent that the Raspberry Pi Foundation has to have income that at least meets all of it's costs (that is, does not run at a loss), it is like any other commercial entity. The Raspberry Pi Trading Company is a conventional commercial business. The RPT is a wholly owned subsidiary of the RPF and all of it's "profit" goes to the RPF.

The primary characteristic that distinguishes a charity or other "not for profit" organization is that it cannot use its net income to enrich individuals. Those funds have to be retained and used for its declared purpose. In the case of the RPF, that purpose is education. And--by the way--it is the RPT that funds and sells the Pi0 and Pi0W.

As for the general contention that production of the Pi0/Pi0W could expand endlessly, that fails on two accounts. The first being that the capital to fund production comes from the RPT, a fairly small company. If they were to borrow funds to greatly expand production, the cost of borrowing the capital would eat away the limited profit that the RPT makes and it would *become* a "loss leader". The second problem is really manufacturing capacity. Right now, so far as any of here know, the Pi0 and Pi0W are made when the production line would be otherwise idle, with all pending orders filled, or when switching to another product isn't practical because it's too close the quitting time for the people working the line. One certainly wouldn't want to use staff overtime to make Pi0/Pi0W boards. The extra labor cost would--again--turn a small profit into a decided loss. Likewise, one doesn't want to take time away from making boards with a significantly greater margin, as that would eat away at the overall profit for the line. As a result of all this, the Pi0 and Pi0W boards are made in the "interstices" of the main production work and any other approach would turn a slim profit into a loss.

So the "TL;DR" version is, no, production cannot be economically scaled up beyond fairly modest levels.

In the end, those who want to buy multiple boards will simply have to wait until the market is close enough to saturation that demand drops off because there are legitimate supply constraints.

Return to “General discussion”