Guest Blog: Industry 4.0 — Reinventing the Factory Stack by Robin Dechant

If you went to bed last night as an industrial company, you’re going to wake up today as a software and analytics company.

Jeffrey Immelt, former CEO of General Electric, nails it down. Industrial companies are feeling the increasing pressure to adapt to the changing environment. They experience the shift to new business models and the need to implement new technologies. After steam, electricity and computers, there are now a huge wave of new technologies shaping the industrial sector, broadly described as cyber physical systems. This includes cloud computing, IoT and bringing together the physical, digital and biological worlds.

In the following post, I want to outline my thinking around evaluating Industry 4.0 startups, present opportunities and risks of this market, map 100+ companies as examples and try to give some advice to founders in this industry based on all the discussion I had. For a broader overview of the different industrial revolutions, see this Forbes article, Bastian Bergmann’s post here or watch this video from the World Economic Forum.

The four Industrial Revolutions (Source: Marcellus Drilling)

New concepts and technologies are evolving

In the last few years, several new concepts have evolved that are changing industrial value creation: From the R&D phase to the manufacturing and assembling processes, until shipping the product to the end customers. Some of the most interesting ones are:

Lights out factory: When I first heard about this concept, I was actually surprised that the first lights-out factory has been operating since the1980’s. The term is basically a synonym for an autonomous factory, meaning that the factory is running 24/7 without any humans involved.
Cobots: With the pace of advancement in AI, many people talk about which jobs will be replaced by robots. Instead of machine vs. human, Cobots leverage the collaboration of both. These robots are often highly adaptable and can support humans in repetitive jobs. Interesting examples include Festo’s ExoHand or Skoda’s factory where cobots support workers at the mechatronic assembly line.
In-memory computing: This is important for IoT to collect and analyze data in-memory on a single data copy on platforms such as SAP HANA.
Edge computing: coined by Cisco, edge computing allows companies to process data as close as possible near the data source and not in the cloud. Advantages are less latency of transmitting the data to the cloud and higher safety.
Machine Learning and AI: This one is obvious. Countless data streams along the whole production process build a solid basis for gaining predictive insights that go far beyond any traditional Manufacturing Execution System (MES).

Shift to service based business model

There are more and more signs that products in the industrial world are being sold as products-as-a-service or solution-as-a-service rather than as standalone products. This sounds familiar if we look at the shift from on-premise Software to SaaS. Why does this make sense and what are the advantages in the industrial world?

For Sellers: like SaaS, companies can benefit from an increased customer lifetime value and a lower barrier to entry. In addition, they can generate revenue even in crisis, if nobody wants to invest in new machines or assets.
For Buyers: they benefit from higher convenience and more services such as predictive maintenance or condition monitoring.

**Product vs. Service orientated Business Model**

This shift is happening globally, but independently from this trend it gets harder to differentiate by quality for western industrial companies, since manufacturing companies in China and elsewhere are catching up. This leads to the question of how to differentiate and win over the competition that can usually produce at cheaper costs?

Customer responsiveness that allows a faster time to market and to adapt faster to changed demand. Customer focus and individualization to increase the variety of products are also a good idea.
End-to-end solutions along the whole process from R&D to After-Sales. Build interfaces, APIs, share data with suppliers and customers to work collaboratively on more efficient processes and to reduce costs and time.
Leverage automation and robotics to achieve similar labour cost as in low income countries.

The Factory stack is different from the Software stack

For investors that traditionally have invested into software companies selling to other software companies or companies that use an increasingly amount of cloud-based products, investing in the industrial sector is a bit different. Self-service applications that we are used to from SaaS companies that sell to prosumers, freelancers or SMBs are very rare to non-existent in the industrial world. It is usually an Enterprise-sale that makes it hard for early stage investors to pick the winner at the seed stage. Moreover, the products are rather complex and often not intuitive to everybody without much knowledge about the industry.

The factory stack usually consists of different machines, different sensors and different transmitters that make it hard to build a standardized plug & play solution for various companies. A good way of summarizing that comes from a discussion I had with an Entrepreneur working in this field:

“After you have visited 10 different factories you think you have seen everything, but then you go into the 11th factory and it’s completely different again.”

I believe that industry 4.0 is not really about replacing machines and equipment but about leveraging software, exploiting the captured data and making machines and human workers smarter and more efficient. Instead of replacing machines, manufacturers might add some sensors but the true value will come from the software. Think of updating your machine through the cloud similar to a Tesla. Now is the time to reinvent the factory stack.

Obviously this doesn’t come without some risks:

Cybersecurity risk: malware to attack industrial automation, to gain control over devices or corporate espionage to steal sensitive data from competitors.
Production downtime: software failure or no connection to the cloud that will lead to stopped production. In the automotive industry, one minute of production downtime costs $22k (!)
Quality losses e.g. for new inspection software that is not as accurate as the previous one at the beginning but it learns over time (AI software that learns over time).
Interoperability: little technology standards that make integration and interoperability with lots of different devices very difficult.

Applications after process steps

What I have written above lead me to the question: which are some of the most interesting opportunities in the industrial sector? In the following, you will find an overview of some companies in the industrial sector that I was looking at for P9 and for general interest in the field. It is by no means a complete landscape. If you feel that your company should be included, please reach out here.

Engineering Tools: 3D modeling, prototyping tools and simulation platforms for designing products.
Prototyping / Finding Supply: rapid prototyping with the help of 3D printing, platforms to find suitable suppliers and vertically integrated factories such as Plethora.
IoT / Middleware: get data from machines, connect offline devices with online services. Connected devices that are able to collect and share data that can be used for real-time monitoring or further analysis.
Shopfloor Guidance / Apps: enhance work instructions for complex processes, process security and to ensure the quality of production. Often focus on smartphone, tablet and modular workstations.
Robotics: software to program robotic behavior, AGVs and other sorts of robots. Investments in robotics have taken off recently.
Wearables: touch interfaces are ubiquitous in B2C and people are used to personal devices. This trend is recognizable in the industrial world.
Analytics / Efficiency: for a 360° overview and full control of the whole production process. Measure and analyze human workers and machine work on the shopfloor including condition monitoring of machines andenergy consumption.
Inspection: companies that help in discovering issues on the assembly line, e.g. with the help of computer vision.
Predictive Maintenance: Solutions for condition monitoring, optimizing performance and reducing downtime.
Asset Tracking / Location Analytics: get transparency across the supply chain with the help of tracking devices and predictive / prescriptive analysis.

Incumbents are not as asleep as one might think

Let’s take Germany as an example where 23% of the GDP value add comes from manufacturing and where 48% of the mid-sized world market leaders — the so called Hidden Champions — are coming from.

While it’s true that they may not take that much risk and not invest in new projects as heavily as the GAFA do, they do put effort into digital initiatives and adapt their business model. One could argue that if it’s about incremental improvements, incumbents will do it — if it’s disruptive or 10x better products, incumbents might go too slow. Here are some examples:

Kärcher — moving cloud-first: they are working together with AWS since 2012. Their cleaning machines have a telematics box that sends machine data to the cloud such as the location for more efficient planning and management of maintenance services.
Viessmann — a healthy risk appetite: The heating and refrigeration manufacturer has its own VC fund, a company builder based in Berlin and tries to create community around IoT with Maschinenraum. The whole company is experimenting a lot with new business models and ideas and is one of the most forward thinking German Mittelstand companies in my eyes.
Kaeser — changing its business model: The air compressors manufacturer put sensors in its compressors a few years ago and thereby changed its business model from selling compressor machines to selling air-as-a-service. Now, customers have only to pay for the amount of air they need.
BMW — the automated factory: The factory for the BMW i3 in Leipzig is quite advanced and has a high degree of automation. This video gives you a glimpse.

And they have to be active. It is probably going to be much easier for software companies to enter new industries (e.g. Google → Automotive) than for traditional industrial companies to hire top notch developers.

Implications for founders

Obviously, this whole development opens up a huge opportunity window for Entrepreneurs who want to transform the industrial sector. Some things I would recommend to keep in mind:

Customer focus: work closely together with customers and pilots from the very beginning. Develop the product based on their feedback, try to have short iteration cycles. It is definitely okay to spend time with them if they use the product and give you feedback. Compared to SaaS companies that sell to other software companies you cannot A/B test. Make it easy for them to try your solution, e.g. start with one production line instead of the whole shop floor.
Avoid non-paid pilots: my impression is that the bar for doing a pilot is fairly low. A lot of companies are willing to test your solution, but often they don’t want to pay for the pilot. I know it’s sometimes painful but have the confidence to say no. There are several enterprise SaaS companies that could grow from their own once they close the first enterprise deal — aka hunted the first elephant (see my post here). Additionally focus on one or two use cases for the pilots instead of having a pipeline full of small pilots of different use cases.
Sell a use case: sell a clear use case that people from the industry understand easily. Instead of selling a “dashboard”, sell them a “control room”. Adapt your language to the industry to your best understanding and sell ROI-first.
Try to sell high: it’s good to talk to the workers in the R&D department or on the assembly line but in many cases try to sell the product as high as possible. Call C-level management, the Head of Production Planning or Director of Manufacturing. The worst thing that can happen is that they forward you to somebody down the hierarchy.
Understand enterprise sale: try to understand the enterprise sales process. Who is the user and who is the decision maker? Who has the budget? How does the procurement process look like?
Platform-second: instead of trying to build a platform first, try to start with a narrow use case and develop the product with the goal of having a platform in the long run, especially for IoT. People don’t buy IoT, they buy a solution to a problem.
Avoid aimless pitching: due to the increased interest from the industry, many corporates invite startups to pitch and there are lots of events with possibilities for pitching. Think again if your time is well spent there before accepting. Many times it is about one-sided knowledge transfer rather than the interest in funding or collaboration.

These are just a few learning from previous conversations I had and I’m more than happy to talk to more industry experts and Entrepreneurs in this field. If you want to change the industrial sector, I would be very happy to hear from you.

Many thanks to all the inspiring conversations I had in recent times with entrepreneurs, industry experts and fellow investors.

Robin Dechant is an investor with Point Nine Capital, an early-stage venture capital firm primarily focused on SaaS and online marketplaces. He spends a lot of time looking at the manufacturing sector and is interested in companies that are trying to reinvent the factory stack. Here's a link to his monthly Manufacturing the Future NL.