Industry 4.0 is a business model revolution, not a technological revolution

Allow us to introduce a tech company within a listed industrial blue-chip: Applied Manufacturing Solutions is an in-house software business in Sandvik's Machining Solutions business area. In our latest Nordea On Your Mind, 'Industry 4.0', I talk to Hugo Nordell, Vice President of Strategy & Business Development, and Ulf Hermansson, Director of Strategic Execution, about digitalisation of Sandvik's business, what AMT does, the hype around Industry 4.0 and the potential shift from product sales to usage-based business models.

Could you tell us a bit about Sandvik Applied Manufacturing Solutions, and your respective roles in the Sandvik group?

Hugo: At Sandvik, we have a fundamental belief that software is the new value driver in manufacturing. Everything we do in Applied Manufacturing Solutions comes back to this one statement. The statement is both an opinion and a description of what we see happening out there in reality. We see that our customers, manufacturers of all shapes and sizes around the world, want to move towards a much more flexible production setup. And this is because they want to address their cost base, which is currently very rigid and driven by labour, materials and overhead. In addition, many manufacturers do not have very high profit margins.

Customers want to digitise their own feedback loops. The direction of the trend is the same across advanced economies, developing economies and China. Productivity growth in the economy and in manufacturing industries is stalling. Customers are used to annual productivity growth rates of 2-4%. The size of the total market addressed by Sandvik Machining Solutions is several hundred billion USD. If the players in this market target annual productivity growth of 3% and were to achieve this over a ten-year period, it would represent cost savings of more than USD 100bn. A critical tool for achieving this is what we call engineering software.

It has already been shown that software can substantially reduce costs in other industries. The same is true for manufacturing. And engineering software can both benefit Sandvik's own cost base and our revenue streams as well as their volatility. Applied Manufacturing Solutions was created both to tap into the countercyclical characteristics of engineering software and to build a business from engineering software. We see a lot of synergies from our own experience in manufacturing and our understanding of our customers' processes and operational reality. We see the potential to drive recurring revenues, which we do not really have in Sandvik Machining Solutions today, which could help reduce earnings volatility.

As you know, Sandvik has three business areas: Mining & Rock Technology, Materials Technology and Machining Solutions. Applied Manufacturing Solutions is part of Machining Solutions, and expands Sandvik's addressable market. We are a pure software business and sit on the same level in the organisational hierarchy as the tooling divisions Coromant, Seco Tools, Walter Tools and Dormer Pramet. We have our own budget and P&L, and we are entirely business accountable. We focus exclusively on engineering software for component manufacturing, which we sell both internally and externally. We believe we have world-class solutions, and we of course want Sandvik to make use of them as well. Sandvik does not disclose specific financial data for Applied Manufacturing Solutions, but I can tell you that after being officially formed in mid-2018, we are several hundred people today, mainly software engineers. And many of them are designers and data scientists. It is not only about engineering competence and about implementing the manufacturing know-how we have at Sandvik into our software. We think it is critical – particularly if you aim to have a premium product – to offer a compelling user experience and make your customer feel empowered. Understanding how our customers behave allows us to build more attractive solutions.

The value chains of our customers typically have four phases, where you can describe each through the question they need to answer for that particular one:

1. Design and planning – 'How do I make this?'
2. Preparation – 'Is everything in place?'
3. Execution – 'Is everything running right?'
4. Evaluation and verification – 'Did I hit my end customer's specification?'

In Sandvik Machining Solutions, anything that is not actual metal cutting is adjacent, so that is where Applied Manufacturing Solutions operates. We are active both upstream, where a lot of our customers' production decisions are made, and downstream, where we can actually be part of the control point where customers determine the quality of their production output. We can help customers make better choices early, to avoid big swings in quality at the end of the process. We want to change this decision-making from a straight line to a closed feedback loop. If today's typically exclusive human operator decision-making for all four phases is replaced by such a loop, we will be able to add a fifth phase and answer its question of 'How do I improve?' And this is at the heart of the lean production philosophy. With this in mind, I can give you some examples of what we offer today in Applied Manufacturing Solutions.

• We have a costing and quoting software service solution, currently run as an open beta, called uFab, for the design and planning phase. Our uFab has an artificial intelligence component, which tries to figure out a rough way in how you could approach manufacturing, which machines to use, which tools to use, the production strategy, etc. A human operator can use this input to refine ideas for how to produce and what to quote, and potentially improve the odds of winning an order just from being faster to quote than anyone else.
• Then there is Prism, which operates a bit in design and planning, but mainly on the shop floor. It is a solution for programming metal-cutting machines that customers use to manufacture components. In Prism, we have created a cloud-based iPad application, which is point-and-click. You use your finger to navigate a 3D model of a component you want to produce, and you simply tap on an area you want to machine. It has been live for a few months.
• CribWise is shopfloor logistics, tracking the quantity and location of tools, and availability of safety glasses, safety shoes, jackets, gloves, batteries, etc. It helps customers address machine utilisation, which can be improved up to 20%, and working capital levels. We have seen customers who have tools with a value corresponding to 10% of sales which they do not use in inventory, and they are not even aware that they have them. CribWise also automates procurement of consumables, according to thresholds that customers specify.
• TDM Systems is a solution for digital lifecycle management of your tool inventory. Customers typically have thousands of tools from six to seven suppliers, which is challenging to keep track of. TDM lets customers know what tools they have, where those tools are and what their cutting lifetime is. And it also manages the digital twins of your tools, so they can be used effectively in the design and planning of manufacturing.
• In the evaluation and verification phase, we have two solutions. One, by a company called ATT Metrology, is an engineering service business working with the largest aerospace and automotive manufacturers to design their quality inspection processes, typically for entirely new products. They help determine how quality for a new aircraft body or new car model should be measured, and how the process for measuring that quality should be automated.
• The other company is called Metrologic Group, which was our first major acquisition. It is based in France, and builds 3D quality inspection software. It is used to run measurement hardware based on touch, either a probe that contacts a surface, or a laser that measures it. Their software can be used with measuring equipment from different vendors, which means staff only need to learn one system.

If we simplify drastically, and call Industry 1.0 basic industrialisation with steam engines, Industry 2.0 electrification, Industry 3.0 programmable logic from the late 1960s, and the currently hyped Industry 4.0 ‘communication between people, services and things’, what opportunities do you think industry 4.0 could generally offer manufacturing industries, and what are the key enabling technologies?

Hugo: I have a very crude view of Industry 4.0. Every industrial revolution has provided us with a fundamentally new set of capabilities to drive productivity growth. To my mind, this is what it is all about – doing more with less. It does not matter if it is labour productivity, capital efficiency or minimising your environmental footprint. It needs to be about productivity growth to matter to manufacturers. And to be perfectly blunt, there are surprisingly few killer applications for industrial productivity growth out there today.

When new markets are developed, there is typically one company that moves in and figures out the killer use case to revolutionise the customer experience, like the first iPhone in 2007. This company typically realises that every single existing solution simply does not cut it, and ends up building most of the new solution itself. Hence, the first successful iteration of any new market is typically a vertically integrated solution. This is usually enough to get the early adopters and part of the early majority to sign up. As the market grows and is proven, new entrants arrive, who start modularising the solution stack. This is what happened in the semiconductor industry, for example. The vertically integrated pioneer usually does not dominate the new market, but there is room for highly specialised niche players in different parts of the value chain. Part of this comes from customers over time learning to be much more proficient in specifying their requirements from each part in the value chain.

Looking at Industry 4.0 or the Internet of Things, or a mix of the two, in a historical context, I can think of perhaps one or two applications that really blow my mind. And this is after probably over USD 100bn of investment in this area in the past six to seven years. I think this poor result partly owes to new entrants putting money into it expecting to own the entire stack and build their own walled gardens, wanting to own every control point with the customer, arguably like the App Store for iPhone. The problem with this approach is that you are projecting onto an industrial manufacturer the mindset and purchasing power of an end consumer such as a teenager who wants to download the next Candy Crush Saga from the App Store. A big manufacturer might have billions in revenue and thousands of staff, and a completely different bargaining power compared with a teenage consumer. Looking at early IoT successes like PTC ThingWorx or GE Digital, they predicated being to industry what App Store has been for smartphones. But the dynamic does not hold, because customers in this space are far too proficient and have far too much purchasing power to let one supplier dominate. There is no Wal-Mart effect to be had. They are terrified of having a single supplier.

It is only in the past 18 months or so that we have started to see a meaningful change in attitudes. Players who wanted to do it all alone are now returning and asking customers to partner with them. This has allowed ecosystems and partnerships to emerge and thrive when vendors have become prepared to share value with customers.

To my mind, a lot of the enabling technology is already there. A decent-sized manufacturing facility today has internet connectivity, high-performing local area and wide area networks, and machinery which is already sensorised. There are systems tool kits and and open APIs making it possible for machines to communicate. What is missing are business incentives and ecosystems allowing the different participants to have aligned interests. I would make this analogy, which also highlights why I am not so bullish on 5G mobile networks: We do not need bigger highways with more lanes and higher speed limits. What good will they do if none of us have any cars to drive on them? Industry does not need highways, it needs cars.

What do you think are the main obstacles to implementation of Industry 4.0? Do you see big differences between industries?

What you need to get started is business alignment, and understanding of what KPIs you want to move and how you could move them. How do we drive above-and-beyond business outcomes? Today's situation is what you get if you hand responsibility for what is a fundamental business change to your IT organisation. This change should be driven by a company's top leadership. The change has to be designed business first, with whatever technology you choose being a result of what business outcomes you are looking for.

I am not convinced there is a technology paradigm that justifies us labelling it as 'Industry 4.0'. This is what people want it to be, but I have yet to see proof that it is. The proliferation of the internet has put us in the midst of a profound change in how information is shared. It is the highway. It already exists. I would argue it is somewhere between Industry 3.0 and 4.0. Industry 4.0 needs to be business model innovation. Machinery is already very sophisticated. When a typical customer buys a machine tool, they buy a Lamborghini. But they drive it like a Volkswagen. So there is massive under-utilisation of capability out there. What will allow customers to address this is a sea change in business models, which incentivises participants in ecosystems to contribute to a better use of this capability. And such participants could include non-industrial players like software providers, financial institutions, underwriters – take your pick. The customer owns the assets, but uses input from all these players to get more out of them. To me, this is what Industry 4.0 is all about.

Digitalisation could offer alternatives to traditional manufacturing business models such as ‘produce-sell-invoice-get paid-repeat’; Do you expect to see emerging usage-based business models for industrial equipment? What could be their benefits compared with traditional models? Could there be any disadvantages? Are non-industrial partners needed to enable new business models?

Hugo: Absolutely, there are some companies trying this out already. There are some striking examples of such business models being great successes, like cloud infrastructure providers. The big three, Amazon Web Services, Microsoft and Google Cloud, all operate this already, and it is a huge market. When I refer to an atomic unit, I refer to the smallest common denominator that you can define, which allows customers to achieve something that matters to them. In the case of cloud services, it is a computing cycle. They have completely reimagined the way you can consume that, and how you pay for it, through managed services instead of buying a computer. It has evolved so rapidly that in many markets it is today completely alien to even imagine running your own server rack. The cloud service providers figured out what their atomic unit is, and industrial manufacturers need to figure out what their equivalent is.

For big players who already run a major aftermarket business with spare parts and service, I imagine this will be easier. And it will make sense for big costly equipment which represents a major investment, but less so for small-ticket items like tools, which can have an economic life of less than a day. In manufacturing, equipment for mining or oil and gas exploration might for example be more applicable as a starting point for usage-based business models. There, the amount of material excavated is a key KPI, for which the equipment used to extract it is the critical driver. General manufacturing is more messy, with a vast number of things on the shop floor which can influence output, of which the machine or tool is just one. If we tried to sell a level of scrap rate to be achieved by a customers, there are probably hundreds of factors which could affect it beyond the choice of tool. So the current opportunity for introducing usage-based business models depends greatly on the vertical, and what it is you are selling.

Usage-based models are not devoid of risk. As any enterprise which has moved to cloud-based computing knows, if you do not understand what you are actually paying for, you could end up paying way more for cloud services than what it would cost to run servers locally. Especially if you do what is called 'lift-and-shift', taking a platform designed in the 1990s and run it in a managed services cloud environment. The same is true for large equipment manufacturers and their customers. A transition to a usage-based business model requires preparation and communication to make the organisation and its stakeholders aware and ready for the big change to revenue and cash flow dynamics from replacing equipment sales with recurring subscription revenues. This is a big challenge. You can think of it as 'the valley of death'. Many will likely begin this journey, but not all will make it through. If you have a new business model, but your organisation is not structured to enable it, then that is what kills it.

Usage-based business models may become widespread, but not universal. You would be surprised how many customers want to own their equipment. There is also a risk of overreach. Say that you supply one small, but critical, component, and you seek to offer what is essentially a full business process process outsourcing. Will you actually be able to do it? Maybe. But even then, there is the question does the customer want you to do it? Not necessarily. If we were to sell tonnage of rock crushed as a service, we could know how big their crushed tonnage is, and figure out what their productivity is, and hence how much money they are making. Customers may not want their suppliers to know how much money they are making. These kinds of misalignment are surprising common when you start scratching the surface of usage-based models. In cloud services, it is working because the atomic unit is very generic, applicable for everyone. This is not necessarily the case for industrial equipment. 

Ulf: Ten years ago, I was responsible for business intelligence at Sandvik. At that time we relied heavily on the Porter model, and everything was about competitor intelligence. They were our competition, which we also looked to for inspiration on how to do things even better. Today we look more at pioneers, new entrants who can break through entry barriers. It is no longer as relevant to look at traditional supply chains with materials, components, finished products and end customers. Technologies in communications and electronics have been around for many years, and in the past ten years we have seen the consequences in terms of traditional business models starting to be challenged. The context and focus is shifting from competitors to ecosystems.

During the industrialisation era, when the steam engine was introduced and later, after electrification, the concept of machining was developed and it attracted top talent. The top people wanted to work with that. A new community, the metalworking guys, was formed. Several decades later, the first numerically controlled machine tools appeared, and that community was split into the traditional old school viewing the work as craftsmanship, and the pioneers seeing the potential from combining traditional skills and equipment with computer power, which offered exponential optimisation potential as opposed to incremental improvement for traditional approaches. This has been going on for several decades, and led to a situation with machine shops today often competing for talent and whose alternatives are simple, perhaps minimum-wage jobs. This is putting immense pressure on manufacturing industries to change, for example to offer knowledge as a service, and to grow productivity with the help of software and new business models, rather than through better machines and craftsmanship. Even at my age, I have come to realise the power of the transparency that the internet offers. I used to believe that you will always turn to your supplier for help and advice. But today you can compare different options quickly and easily online, and make your choice. This transparency is hitting industry hard now, eliminating pockets of excess returns. And it brings pressure to introduce new business models that can attract and retain - and perhaps require – top talent, which can thrive under these circumstances.

What digitalisation priorities and initiatives are Sandvik pursuing today? Could you give some examples of projects in Sandvik Applied Manufacturing Solutions or elsewhere in the group?

Hugo: All our digital initiatives are oriented around Sandvik continuing to be seen as a premium provider, ensuring that we have superior customer relevance and continue to our customers' total productivity partner, also in the future. Ultimately, that is how we avoid commoditisation and competing purely on price. It is also important for being able to attract young talent. Our industry has a rapidly ageing workforce. In the US, 15% of the 12.5 million-strong manufacturing workforce should already have retired. They represent a huge potential skill shortage. And at the same time, there is a war for talent with new skills and qualifications. They do not choose between us and our traditional competitors when looking for a job. They choose between us and a tech company or a startup. This is why I think employer branding will be much, much more important going forward than it has been in the past.

Ulf: Agility is a key word in this context. It could apply to individual projects, where people realise we need to take a decision, we need to be fast. Another key area is data, internal and external. The use of data for our ability to react and respond to external signals, such as adapting our output quickly to changes in demand, can dramatically impact our flexibility. If we can read the signals quickly and be able to respond quickly, we reduce the vulnerability of our profits to changes in demand. Agility figures in conversations everywhere across the Sandvik group, from salespeople to product manager all the way to the board room. We have lots of data, and it is a matter of how we collect it, process it, analyse it and act upon it.

Hugo: Another focus is how do we dramatically shorten and automate as much as possible from order to delivery. We have many initiatives to enable that. And there is of course the establishment of a software business, Applied Manufacturing Solutions, within the Machining Solutions business area, as its nature is completely different from producing hardware.