Create impact with Industrial AI!

Industrial use of AI is not an end unto itself. To be successful, AI applications must have far-reaching impacts on competitiveness and offer tangible added value – for society as a whole, and for all sectors of industry – in terms of productivity, innovation capacities and sustainability. To achieve this, the needs and expectations of European industry must be aligned from the beginning, across the entire supply chain (e.g. component suppliers, machinery manufacturers, IT companies, OEMs). AI offers Europe a unique opportunity to achieve technological leadership. By rapidly integrating AI into their offerings, European industrial players can increase their competitiveness and profitability.

Four components drive AI innovation:

1. AI models
2. Access to data
3. Cloud capacity and high-performance computing infrastructure
4. Access to talent

Europe must become more agile and improve in all four areas! Let’s focus on the manufacturing industry and some examples like design and engineering – The same principles apply to other applications and sectors as well.

A major challenge in terms of ensuring the competitiveness of the manufacturing industry is optimization of products design and engineering and production systems. Presently, design and engineering are characterized by significant specific effort with little reuse from previous projects. Consequently, design and engineering projects are often risky.

Increasing time-to-market pressure requires reductions in processing time for design and engineering projects. One approach to achieving higher quality engineering results in early project phases is clever reuse of available data from other, comparable projects, not only from projects within one’s own organization, but also across companies. Building on proven methods reduces risk, costs. and processing time simultaneously. AI offers such clever reuse. Siemens, cooperating with industrial partners, has launched an Engineering Copilot.

Embedded in its engineering tools, it is a Generative AI tool designed to assist engineers in generating automation code, creating visualizations, and performing natural language document searches. Engineering Copilot aims to reduce workload, increase productivity, and address labor shortages by augmenting engineering processes with AI. The Copilot, as part of Siemens Xcelerator, is available in the Xcelerator marketplace. Initial pilot tests indicate significant time savings and error reduction in industrial automation environments.

Building a cutting-edge industrial AI application, such as the Engineering Copilot and related solutions, requires a collaborative approach where data and knowledge can be shared in a secure and trusted environment. It requires data and knowledge, often across different industries and sectors, as well as talent, experience, powerful infrastructure, and lean efficient processes.

In such an environment, interdisciplinary teams of experts collaborate, motivate and inspire one another. Additionally, shared infrastructure is readily available for development and maintenance of AI solutions. Protecting competition-relevant knowledge and intellectual property rights – for example production design and automation code – is a key horizontal requirement. One must credibly ensure that the parties involved agree to share data and cannot leak it. This must be reflected in cybersecurity and legal requirements (e.g. related to data protection, GDPR, ePrivacy), but also in contractual regulations. Well-maintained data management, e.g. in and between companies that work together in data ecosystems, is another key prerequisite. Data must be accessible, structured, semantically described and embedded in a secure and privacy-preserving environment. As a result, in many areas today, AI applications require an upstream ETL process (Extract-Transform-Load) and technologies such as federated learning to meet these requirements. On the shop floor, each machine generates its own data so that AI can be used profitably. This requires semantic and syntactic standardisation on all levels (from machine level up to the company level). International standards for interoperability are available on the market today, are being used, and will be further improved. Finally, it is crucial that the barriers to enter are low for all contributors and all partners such as companies (including SMEs) have broad access and can easily and efficiently create effective business applications. To achieve this, industry associations such as VDMA (example from the manufacturing domain) must play an important role.

Europe and the European Member States have several relevant initiatives, if properly aligned, can become important building blocks for this European AI ecosystem. With the “EuroHPC” Joint Undertaking, we have the elements for expansion and development of High-Performance Computing (HPC) in Europe relevant to have the necessary computing and storage capacities – a super compute infrastructure – to foster and support innovation in areas such as AI. This includes infrastructure, providing the necessary capacity and secure environment for the computation of complex applications (e.g. simulation, prediction and computation of AI models) and supporting usage of that infrastructure. “AI Factories” create a dynamic ecosystem fostering innovation and development in the AI field and play a central role in strengthening Europe's leadership in trustworthy Industrial AI. AI Factories – the open AI development ecosystem formed around European public supercomputers – have the potential to provide a collaborative environment by leveraging strong interdisciplinary AI ecosystems across Europe. These AI Factories integrate high-performance-computing, access to data and expertise to support SMEs, startups, research and innovation ecosystems, and other in the development and training of large AI models.

The “IPCEI CIS” (Important Projects of Common European Interest on next generation Cloud Infrastructure and Services) aims to develop an advanced cloud infrastructure, new edge capabilities, and cutting-edge cloud services in Europe. It promotes investment in innovative digital infrastructure for cloud-edge and connectivity technologies to drive the digital transformation of the economy.

The cloud edge continuum supported by the IPCEI CIS describes the seamless connection of edge computing and cloud infrastructures. This makes it possible to process data closer to the source. while taking advantage of the cloud, to ensure efficient and scalable data processing, which is particularly important for IoT and real-time critical applications. Europeans leading “Data Ecosystem initiatives” such as Manufacturing-X initiative, are driving key concepts for the further development of data ecosystems. To accelerate the digital transformation, the industry seeks to increase efficiency and flexibility in production and improve sustainability, resilience and competitiveness with benefit from data ecosystems – the data ecosystem part. Technologies such as artificial intelligence (e.g. AI Factories – AI development ecosystem) and cloud edge computing (e.g. IPCEI CIS – digital infrastructure), used in combination, processed in computer systems (e.g. EuroHPC – supercomputer infrastructure), create impact with Industrial AI while ensuring data sovereignty. Combined with data ecosystems (such as Manufacturing-X), this enables closer collaboration between suppliers, manufacturers and customers. Ultimately, this increases innovation and competitiveness.

What should be done next?

Greater industry involvement in AI factories at all levels: from start-ups to SMEs and large companies, taking into account the entire value chain. Industry also requires simplicity, scalability, predictability, quality assurance, safety, and sustainability. It is crucial that all AI Factories partner closely to make offerings accessible across Europe for all sectors and partners. It is key to create a networking layer that homogenizes AI Factories at user level, with common governance rules that ensure fair and seamless access to services for industries required to fully achieve their objectives. Finally, as a pre-requisite, a strong supporting scheme between the European Data Ecosystems, AI Factories, super-compute infrastructure, and other
important projects (e.g IPCEI CIS) should be established in close cooperation with Industry. European Associations such as BDVA, a private member of the EuroHPC Joint Undertaking and focused on data, data value and industrial AI, are fundamental for effective engagement of Industrial AI, dialog with policy making, and among AI factories. This will foster collaboration between the Data ecosystems, other infrastructure projects, and involvement of Industry Associations.