This is a regular discussion I have with prospect or with different business oriented people about producing electronics in developed countries and costs of devices. When you talk about electronic products in many minds it is equivalent to talk about Chinese production & design. It’s like when you talk about software in a couple of years ago it was a kind of synonym  with back-office Indian development.

For sure these short cut have a part of reality for some main streams product because they are main stream and because you need for them a workforce you can only find in China / India.

In my domain, I’m not making mainstream and people I’m talking with are not either. Making a hundred even a Thousand of IoT device in China it’s like if you want to make your personal website done by a large scale service company located in India. I’m quite sure you will pay a lot, never been really delivered and at least get something that is not what you order. This is due to distance, languages, processes and skills (not adapted to create personal webpages).

This is to illustrate in IoT you have different kind of project like in software and the tool (factories, design…) you need differ every-time. To illustrate I will explain what are the cost for producing IoT.

Understanding IoT cost is really important if you have an IoT project : I had this discussion with many clients, maker and contacts who generally have 2 wrongs ideas in mind :

• The electronic you buy from china is costing 5$, not because it has been made in China but because it has been made in MILLION of units.
• The prototype you did on your own for 15$ is not relevant to determine the final price of the product and to sold it as a final product.

In fact the cost of IoT product depends on the size of the batch you are producing (1st effect), the level of automation in the production (2nd effect) and the Bill Of Material (BOM) cost (3rd effect) at end.

Size of batch you are producing

The number of board produced is really the main important point. The global number of product is a key as whatever the number of board you make you have a lot of fixed costs we can estimate like this :

• Board design / prototypes / iterations – from 1k€ to 20k€
• Certifications – from 10k€ to 30k€
• Mold for custom casing – from 10k€ to 50k€
• Production initialization costs – from 1k€ to 10k€

You can do a part yourself, you can choose to use a standard box but basically your global cost will start from 15k€ to 100k€. Generally what you invest depends on the number of product you want to produce. So let’s project this on the unit cost of your product assuming we will invest different amount of money in the design regarding the production target:

• 1 000 devices = 15€ / devices
• 5 000 devices = 7€ / devices
• 10 000 devices = 4,5€ /devices
• 50 000 devices = 1,6€ / devices
• 100 000 devices = 1,1€ /devices
• 500 000 devices = 0,5€ /devices

This is for overall costs. On top of this you have a batch size to consider : in general you won’t make 1000 device on 1 production batch if you plan to sell 1000 devices all along your device life. We can consider you will have at least 10 batch of production before changing your product design.

For each of the batch you need to order components, program the production chain and loose a part of the production, then package and deliver. This have a fixed cost related to the work to do and it also have an impact on the price you pay your components. We can estimate the fix part between 1k€ and 10k€ depends on your batch size. So basically, on top of what we had previously we need to add:

• 1 000 devices = 10 batchs of 100 = 10€ / devices
• 5 000 devices = 10 batchs of 500 = 3€ /devices
• 10 000 devices = 10 batchs of 1000 = 0,5€ / devices
• 50 000 devices = 10 batchs of 5000 = 0,5€ / devices
• 100 000 devices = 10 batchs of 10 000 = 0,4€ /devices
• 500 000 devices = 10 batch of 50 000 = 0,2€ /devices

Level of automation in the production

When you want to produce a batch you need to execute a lot of manual operation, they are depending of the design you have done. Basically to optimize your design for the production you have to invest more in your design. As part of the action you have to consider:

• A wire soldering = 1-2 minutes
• A screw = 10-30 seconds
• Putting the PCB in a box = 10-30 seconds
• Insert a battery = 10-20 seconds
• Uploading a firmware = 10 – 20 seconds
• Execute a test = 30s – 2 minutes
• Add a sticker = 10-20 seconds
• Register in a application = 10 – 20 seconds
• Put in a package = 1-2 minutes
• …

The list of manual operation is large. Lets consider you have a device with a battery holder (2 wires) you need to screw in a box (2 screw) then you need to flash and test, insert 2 batteries, close the box with 2 screws, add a sticker and put in a box. You have a total time of 5 minutes per device. One person will mount 10 device per hour. It will cost you in France something like 1€ / device.

If you want to produce a batch of 100,000 devices you won’t find enough people to do it quickly. For this reason you need to simplify your design and automate the process for this you need to invest in machine :

• Robot for placing, screwing .. 10k-50k
• Test automation 10k – 30k
• Dedicated application for production 10k-20k

So you can have some saving on large scale devices :

• 0 – 10 000 devices = 1€ /device
• 10 000 – 100 000 device = 0,4€ /device
• 100 000 – 500 000 devices = 0,2€ /device

Bill of material

The last part in the bill of material (in my point of view) : I assume you can make really bad choice during the design phase and have it as a major effect but basically you generally make the right choice. Where the bill of material can have a impact is if you select a really low cost solution creating really high cost for design and production when your volume is too small.

As an example : you can choose a SoC with communication modules for your system and have a high cost BOM. For a large number of device it is an error. but for a small number of device it can be a good choice. Let’s put numbers :

• 1000 devices to produce with a BOM of 30€
  • Design is fast
  • Components like SoC and module are tested and certified
  • Your design cost is low : 12k€
  • Production is manual, tests are really low (0,5€) , 95% of production is good
  • Cost per device : 44€

 

• 1000 devices to produce with optimized BOM of 15€
  • Design is complex
  • Certification is complex and you have multiple design iterations
  • Your design cost is high : 30k€
  • Production is complex, test are long (1€) or automation needed, 90% of production is good
  • Cost per device : 47,6€

This becomes totally wrong if you plan to produce 100,000 devices for sure.

Conclusion of this

We can see with these element the impact of the production price regarding the volume of device you want to have.

• 1 000 devices = 26€ / devices + 1,5*(optimized BOM)
• 10 000 devices = 5,4€ / devices + 1,2*(optimized BOM)
• 100 000 devices = 1,7€ /devices + BOM

So let’s take an example to understand what it means : you want to create a product you may basically find on Chinese market produced at million scale at 5€ / device. But this product does not yet exists and you want to start making it. your current capacity to sell it on the market is about 10,000 units. Your production price will be something like :

For 10 000 devices : 5€ means => 5,4€ + 1,2 * 5€ =  11,4€

For 1 000 devices : 5€ means => 33,5€

All what I have written is basically as wrong as true, it is really related to your project and may vary by 80% but it gives the main elements to take into consideration. So please if you have a project take it into consideration.

• Don’t make component price + 20% as public price for low volume
• Stop considering expensive any device targeting niche market
• Don’t be obfuscated when the price Engineers are proposing is growing by two when you divide your target volume by two.
• Don’t ask for making 100 devices at Chinese price.
• Be respectful in front of quickly designed solution with expensive BOM for small batches
• Don’t target custom box design for small batches <=> accept bigger size and be inventive