Exploring Cloud Manufacturing Services

Building products is hard, even for successful and established companies. Despite the challenges, I believe, that over the last ten years it became increasingly easier to build digital products. With platforms like AWS, Azure, Google Cloud and other SaaS tools developers have access to hundreds of digital services, they can validate their idea quickly, and scale it to the whole world. They don’t need to invest in expensive hardware, and they can even get cloud credits for free which last for more than a year.

However, for startups, makers or hobbyists who wish to build physical products, that’s not the case. They can expect a long and painful journey moving from prototype to production and only a few of them are going to succeed. They would need to invest in hardware: buy machines and tools and pay for expensive test equipment. They will probably struggle to find the right skills and expertise and even knowing which process to use would take them weeks. They would have to set up a working facility which includes, usually: a factory floor with various machines; a few work stations; some kind of warehouse with proper asset tracking and inventory/supply chain management. When developing their product they will have to ensure that their product is reliable for the entire product lifetime (under heavy usage, stress, load conditions), that their manufacturing pipeline produces, consistently, high-quality goods. They will also need to verify that their product complies with safety regulation, various environmental conditions (temperature, humidity, vibration, shock, stress), industry standards (CE, UL, FCC…) and recycling and environmental standards (ISO). And finally, if they reach the market and test it, it will take them months to make changes, and iterate.

This is really difficult, and for the last 100 years, big companies like Ikea, Ford, Philips, Nike and Siemens managed to dominate to market due to economy-of-scale advantages, while new entrants could not compete and reach the early adopter stage.

Now, imagine an online SaaS platform, similar to AWS, Azure or Google Cloud, where engineers and industrial designers can design, develop, mass produce and ship physical products. Anything from furniture and electronic gadgets to cars, houses and electrical appliances. This is the vision of Cloud Manufacturing Services.

If you are in the manufacturing space, there is a good chance you’ve already heard about Industry 4.0 and what the cloud is: anyone can run their software in the cloud and store their data there. There is no need to host physical servers and you pay per use. Cloud Manufacturing Services is a different idea, it’s nothing to do with storing or processing data. It’s about using machines like CNC or 3D printers in the cloud and running an entire assembly line there.

A brief history of the CLOUD

I meet many technical people, some of them are experts in their field, who know roughly what the cloud is, but, they have never experienced how simple it is to develop products with it. The best advice that I give them, is just try it!… build something simple, a web app, or just spin up a Linux server. For those of you who don’t even know how to start, I’ll try to picture what it feels like developing digital products today compared to 20 years ago.

20 years ago, when digital engineers (IT) wanted to upgrade their capabilities, for example when they wanted to upgrade their server farm or storage appliance, they had to convince their manager that it is needed, and then the manager of their manager. Afterwards, they had to convince procurement why their preferred supplier is better than its competitors, they might even have to go for a tender process… and then they had to beg finance to allocate budget for it, and then they had to work with legal for a few months to agree on a contract, and finally they had to wait three to six months for the appliance to arrive, whilst trying to free some physical space for it in the meantime, upgrade their electrical and cooling infrastructure and train their stuff… and eventually, they’ve just decided that they won’t. Because it was too damn hard. And innovation stopped. Without IT innovation, the company had to spend more time on manual work and fixing errors and more money on maintaining expensive legacy hardware… If you are working in the manufacturing space, I bet you can relate to this story. Replace the server farm with CNC or 3D printer machine, replace storage with a modern warehouse and sophisticated inventory tracking, and you will understand why manufacturers today struggle with introducing new manufacturing techniques and why developing a new physical product is so hard.

So what do digital engineers do today? They just use the cloud! They have immediate access to hundreds of services and they can scale their infrastructure to support billions of users with a click of a button. They don’t have to worry about upgrades, uptime, or upfront cost. They can set up a whole infrastructure, and delete it after a few hours. And they are doing this while working from home, drinking their favourite hot chocolate.

It’s also worth explaining the revolution that digital technology went through in the last 20 years. As someone who has experience in both spaces, it’s amazing to look back and see how it happened relatively quickly and who did it.

For 35 years IT relied on companies such as IBM, Microsoft and Oracle to supply IT hardware and software to enterprise organisations. It was expensive, prices went up every year, and companies had to pay annual fees of 20% just for standard upgrades and security patches.

And then came Amazon and Salesforce with a different model, what today we call IaaS (Infrastructure as a service) and SaaS (Software as a service). Instead of buying a product, you subscribe to a service. There is no upfront cost, you can invest in innovation and building your product, and the big guys will take care of the rest. And with that, they managed to achieve the ultimate goal: both Lean and Agile!

Why Cloud started with Amazon and Google and not IBM and Microsoft is an interesting topic which I won’t cover here, but there are definitely lessons there for Cloud Manufacturing Services.

Cloud Manufacturing Services

Cloud Manufacturing Services is an idea and a vision. Although there are some manufacturing services which are available online today, they are not integrated and complete. It’s similar to how twenty years ago there were many companies who offered servers and web hosting; it worked and it proved the concept but there is a big leap of service from hosting of servers to the cloud as we know it today.

The best way which I could think of illustrating the vision is by presenting a sample dashboard which includes multiple manufacturing services. When these services are integrated together they can be used to design, develop and manufacture “smart” electronic products (or IoT). The process which I am using here is called Design Fiction, a design practice aimed at exploring and criticising possible futures by creating speculative, and often provocative, scenarios narrated through designed artefacts. Here it is…

As you can see, the dashboard (which resembles a very popular Cloud provider) includes around 50 digitally-defined manufacturing services. When selecting a service, a standard template on how to use them appears:

• Complete input kit definition: What is needed to start the service.
• Variables and options to choose from: i.e. materials, quantity, colour…
• A pricing equation: How much each job is going to cost based on size, weight, quantity, etc…
• Inspection / Test procedure: To verify that the output conforms to the required standards.
• Tag and names to compose this service into software-defined manufacturing pipelines.
• API which implements this functionality and potentially Manufacturing as Code (MaC) script interface.
• Documentation on how to use the service

Let’s go through the dashboard and see an example of how this platform can be used to set up a manufacturing pipeline for a Raspberry Pi kit:

To manufacture the plastic case which is composed of four parts (back, front and two sides), the designer can use either a 3D printing process or an injection moulding one. He or she will upload their 3D files (i.e. G-Code, STL or other) and then for each part, choose the material, manufacturing process, colour and the finish level, potentially leaving some customisation to the end buyer (for example, the end buyer might be able to select the colour of the case). The power supply, the SD card and the heat sink are off-the-shelf components which can be integrated into the kit at any stage.

The electronic board is a bit more complex, The designer would have to upload the GERBER files of the PCB, and the BOM (Bill of materials) for the board. The designer would also need to select the PCB thickness, quality, material and provide assembly instructions for unique components. To test the board after manufacturing and upload the firmware to it, additional information is needed.

To define the manufacturing process for the boxes, the designer would have to choose the paper type, choose the box template (dimensions), and then upload the image for it.

Finally, the designer would need to compose these services into a single process and link it into an e-commerce service.

Implementation approaches

Generally, there are two ways to implement the service:

• Centralised: All the services are closely located within the same factory. (similar to Cloud Compute paradigm)
• Distributed: The actual services are geographically distributed among multiple manufacturers who compete/collaborate to provide the best service.

In terms of feasibility, the centralised approach can start from a spin-off of an existing manufacturer (i.e. Ikea) which is exactly how AWS started. The distributed approach will require an initial investment in shipping/asset tracking infrastructure to support quick and reliable delivery between the various manufacturers.

Summary

Cloud Compute ignited when the internet was reliable enough as a transportation link between users, applications and services. I believe that Cloud Manufacturing can succeed only when there is a good shipping infrastructure that can reliably support the delivery of parts and components between users, manufacturers and product developers. To start the process manufacturers must adopt a standard API and expose their services. Only when there are enough exposed services, platforms like this can emerge.

The interesting question is not whether it will happen, but when, and by who. One way it can start is from companies like CloudNC, 3DHubs, OSHPARK who already provide some online manufacturing services. Alternatively, it can start by big companies like Ikea, Amazon or Tesla who can use their existing manufacturing capabilities. Lastly, it can start from local governments and research institutes who want to increase market competitiveness by leveraging the collaboration and synergy between existing local manufacturers.