Industrial Tech Talk: Predictive Analytics
On 11th of December, Speedinvest i and Aster were delighted to host the “Industrial Tech Talk” at Speedinvest’s rooftop location in Munich. The new event format was set up to bring together some of the most knowledgeable and experienced stakeholders of the industrial tech environment, industry experts, startups and VCs, in order to share thoughts on the topic of “Predictive Analytics”. The idea behind the event was to immerse into the matter and discuss underlying hypotheses and current challenges in this field of industrial tech.
It is well known that Europeans excel at making cars or come up with other innovations, but looking at the startup scene, the amount of venture capital flowing still remains to be significantly smaller, compared to our US friends. Being risk averse, especially German companies frequently feel more comfortable with rather gradual improvements of an existing product than radical innovation. And despite the innovative startup scene, it is a fact that not enough venture capital has been flowing into the industrial sector. This is the reason why Speedinvest i, an industrial tech investment fund, was born. It invests in innovative solutions which revolutionize Europe’s Old Economy.
“Predictive Analytics” as a theme for the first Industrial Tech Talk
There is one significant asset which industrial firms have not yet optimized: their own data. Industrials generate enormous volumes of data, but many companies still fail to make use of this concentration of potential intelligence. Historically, manufacturers have not been the first adopters of new IT capabilities. Nowadays, thanks to significantly cheaper computational power and advanced analytics opportunities, manufacturers are enabled to take advantage of their data to the full extent. Advanced analytics can help industrial companies solve problems which so far, due to lacking data collection, they were not even aware of. The question is clear: how fast will the adoption of new capabilities by industrial players happen?
In addition to the observed transition, the importance of the topic can be backed up by actual numbers as well. If projections for the overall industrial analytics market would be realized, the market will grow from USD 11.29 Billion in 2017 to USD 25.51 Billion by 2022, at a CAGR of 17.7% (Source: marketsandmarkets.com). Industry 4.0 and the introduction of advanced data analytics techniques are the key factors driving growth in this market. When comparing the market size of industrial analytics, the US remains the largest market, closely followed by Europe. However, the European market depicts a greater CAGR — proof of the accelerating need for analytics solutions in Europe.
These projections are understandable considering a wide range of industrial use cases where analytics is advantageous. They extend from maintenance planning and scheduling, failure risk and pattern analysis of assets, anomaly detection, product quality control to process optimization etc..
Discussions and findings during the course of our event
Our event confirmed once again that the most innovative and forward-looking firms in the industry are not ready to use the full potential of their data or do not even have the data available in a structured way. This is really something a company — in probably any industry — should resolve in the first place before even starting any analytics “journey” (in other words digital transformation) at all. Such journeys usually consist of two consecutive parts: the data itself and the analytics part. The latter does not deliver value without having good data on hands.
Speeches by Alexander Thamm (Alexander Thamm) as well as by speakers from our portfolio companies, Michael Baumann (Twaice Technologies)and Alex Appelbe (Metis Labs), took place during the Tech Talk. Each of our guests presented challenging hypotheses about predictive analytics (see below) which later evolved into lively discussions with the audience. Our top takeaways were the following:
I. Alexander Thamm: “Companies first build something and only then scale it — this doesn’t work nowadays anymore”
Alexander, CEO of Munich-based data science consultancy specializing in predictive analytics, triggered discussions around what companies must do to enable predictive analytics in future businesses. He likewise presented his hypothesis that companies must be turned into data-driven organizations with an agile and iterative data strategy, which is closely linked to use case development and implementation. Based on his experience, it is often the case that companies first build a product and only scale it afterwards. In contrast, Alexander assumes that this doesn’t work anymore because “businesses are frequently not plannable”. Furthermore, he supported his view by challenging how some of the companies “want to predict events which have never been measured before in an organization”. “Looking at the data from day one” was his strong call for action for any data-driven company. Concluding his hypothesis, Alexander pointed out that whenever an idea of a startup is B2B, it all comes down to selling!
II. Twaice: “Predictive analysis will never be 100% reliable”
Hands on experience from the startup side was shared with us by one of the founders, Michael of Twaice, a Munich-based startup which provides battery analytics based on digital twins of Li-Ion batteries. Michael’s view on the topic is that predictive analytics will never ensure a fully reliable operation of battery powered products (e.g. in medical applications), since machine learning algorithms are not deterministic and predictive analysis will never be 100% reliable. Therefore, humans will never be fully replaced, specifically in safety critical applications.
III. Metis Labs: “Companies don’t buy AI, they buy value”
Alex Appelbe, CEO & co-founder of Metis Labs (a machine learning platform for process optimization in manufacturing systems) discussed why, with his 10 years of manufacturing background, he is convinced that industrial companies will never be ready to buy AI-first products. “Companies don’t buy AI, they buy value” was the most important message that Alex delivered to the audience: “For success it is important to wrap a technology as a product, while making it easy and simple to buy”.
Industry wishes for ready-to-use solutions — Startups require access to industry
Especially in the industrial sector, it is crucial to rethink how companies work with and find new ways to leverage data (rather than providing basic data bookkeeping services), specifically when implementing IoT solutions. It is vital for most of the companies to make a shift into becoming data-driven businesses. In summary, it becomes clear that the industry wishes for ready-to-use solutions, whether provided by startups or big tech giants. Startups are looking for access to the industry to take advantage of scale effects. Hopefully, the two can meet to benefit together.
You are welcome to join our next event in 2019
The large number of events in the tech-and startup-space demonstrate that the value of data science and predictive analytics is being proven again and again. Our Tech Talk was followed by networking with mulled wine and chestnuts! We thank everyone who has participated in our event and especially our speakers! We will be delighted to welcome you at the next event taking place in spring 2019. Stay tuned and happy holiday season!