Hannover Messe: Focus on (AI) Action in a sustainable way

[chr_hollender]

Data-driven value creation

At Hannover Messe, HPE’s central solution focused on a universal data management project with BMW providing data-driven value creation to the automotive market leader.

BMW Group accelerates their production

The project requires a dynamically scalable edge-to-cloud infrastructure that reliably provides all required data through a data hub. BMW Group has opted for HPE GreenLake and HPE Ezmeral to standardize their data management at the edge and accelerate their production.

Here, HPE GreenLake enables the analysis of data BMW generates worldwide on electric test cars, with the goal of accelerating BMW’s go-to-market. HPE Ezmeral stores vehicle data such as battery temperature, power loss, and vehicle speed, making the data available to BMW Group data scientists and engineers for direct access, regardless of their location. The as-a-service platform also provides a catalog of tools and data processes for analysis and simulation.  On the server side, it runs on HPE Cray (former HPE Apollo) and HPE ProLiant servers in globally distributed micro data centers, along with virtual desktops on the client side. 

Global data availability and management is the basis for any AI innovation. At Hannover Messe, nobody was able to escape the omnipresent topic of artificial intelligence. And frequently it seemed just as inevitable to export internal company data to a hyperscaler cloud to utilize the potential of machine learning and generative AI models. Understandably, this causes some headaches for business decision makers.

AI and robotics, but with data protection

At the HPE booth, however, HPE and German AI innovator Aleph Alpha showed that there is an alternative: Aleph Alpha's generative AI named Luminous has proven that it can stand up to the hyperscalers' GenAI models with additional benefit of explainability – and Luminous can be operated on-prem based on the HPE GreenLake portfolio, which includes powerful AI-native servers. This means, e.g., that a manufacturing company can feed the AI with in-house knowledge and develop its own individual AI model – under full control of the company’s IT department. This avoids concerns about data sovereignty, GDPR compliance, and other regulations.

An example can illuminate the benefits of this approach. A robot manual of several thousand pages is used to train a manufacturer’s individual AI model. This allows staff to ask questions in natural language in case of malfunctions or need for maintenance. In future scenarios, AI could also speed up repurposing a robot for new manufacturing prototypes.

For on-prem AI operations, HPE GreenLake provides an end-to-end platform that enables the development of trustworthy in-house AI applications, such as LLMs (Large Language Models), which avoid hallucinations and deliver consistently reliable results. As the AI is operated locally, this architecture requires no expensive data transfer to the public cloud. This makes the approach suitable even for remote production sites. At the same time, it minimizes risks and misinterpretations (bias), making it easier to introduce AI solutions that comply with the EU AI Act.

Alepha Alpha develops their models in their data center on HPE Cray (former Apollo) systems and using the HPE Machine Learning Development Environment System (MLDE/MLDS) .

Digital Twins optimize industrial processes

In addition to generative AI and AI-powered robotics, another AI use case has enormous potential for optimizing industrial processes: the digital twin.  A digital twin aggregates all the parameters and sensor data of a production process and uses AI analytics of this virtual doppelganger to detect, e.g. trends in component wear or the need for optimization in real time – yet without affecting ongoing manufacturing operations.

To facilitate the introduction of digital twins of production plants, HPE and Bosch Digital Twin Industries (grow platform) presented a joint edge-to-cloud solution for digital twins. Bosch's Digital Twin – IAPM (Integrated Assessment Performance Management) is a dynamic software model of a factory floor or process. It gathers sensor and instrumentation data and analyzes them by using AI. This dynamic, continuously analyzed model of ongoing operations enables a manufacturing company to react quickly to changes, improve manufacturing processes, forecast future scenarios, and base its business decisions on reliable, near-real-time data from its factory floor.

Bosch’s digital twin is powered by the HPE Ezmeral Unified Analytics Platform on HPE servers, which ensure the smooth operation of edge systems even in harsh industrial environments. The offer is already available and comes in standardized “T-shirt“ sizes small, medium, large, extra large, and as a customized solution. Here, too, on-premise operation avoids expensive data transfers and and compliance risks.

How to power the data economy

Data-based value creation ultimately requires data analysis across enterprise boundaries, e.g. across supply chains. Currently, the fact that far too little data is exchanged this way is still a major obstacle that slows down the transition to a true data economy. EU legislators have recognized this and take countermeasures by introducing the EU Data Act: the new law obliges manufacturers of networked products and providers of data services to make data generated within the EU available to users on request, allowing them to share data with third parties. The aim is to make the huge volumes of data that were previously only in the hands of the respective manufacturer or service provider accessible to the data economy.

Initially, industry associations and companies loudly protested that this mandatory data sharing might undermine competitive advantages. However, a survey commissioned by HPE of 400 managers from the German industry revealed a surprisingly positive picture: two thirds of respondents see the Data Act as an opportunity rather than a threat. Respondents see potential mostly in terms of greater efficiency, AI training, and the development of data-driven business models. As expected, they consider the requirement to protect confidential data as the main challenge.

The average data maturity level of the companies surveyed was 2.6 (on a scale of 1 to 5, where 1 stands for data anarchy and 5 for a true data economy) – basically indicating that businesses have already progressed half the way to data-powered business. Nearly half (46%) of those surveyed state that, given the new Data Act, they will rely on a hybrid environment consisting of on-premise IT and external cloud resources – in other words, on the architecture that HPE is championing as well.

Progress for greater sustainability

The transformation to an AI-powered data economy has a critical second dimension: it also needs to be a transformation towards a sustainable, carbon neutral circular economy. HPE, in cooperation with partners, can provide valuable support here, too, as could be seen at Hannover Messe.

HPE is cooperating with Digital Realty, a leading provider of carrier- and cloud-neutral colocation data centers. Based on HPE GreenLake offerings, Digital Realty provides a global data center platform that supports businesses in their AI projects just as efficient as in their sustainability initiatives, such as operating a sustainability dashboard that covers all company facilities around the globe.

And HPE is cooperating with Schneider Electric on data center modernization. Jointly, the two partners help enterprises find their individual pathway to the modern data center by contributing their unique points of view: Schneider Electric, a leading data center infrastructure provider, is well-versed in data center efficiency from power sourcing to cooling technologies, and in capacity scaling. HPE, in turn, knows how to make the best use of available rack space with the state-of-the-art edge-to-cloud portfolio (Intel) servers for regular IT (and some AI) workloads to AI-optimized systems.

HPE partner Iceotope, in turn, combines HPE ProLiant DL380 Intel servers with liquid cooling, resulting in the KUL EXTREME precision edge servers. The plug-and-play-ready systems are optimized for high-performance workloads and low latency. They consume 50% less energy than air-cooled DL380 Intel servers. At the same time, liquid cooling extends the server life cycle and reduces maintenance costs – things bound to shine in any company sustainability dashboard.

Networks for the Data Economy Era

Finally, it is important to note: all these technology trends require fast and reliable networking. HPE Aruba Networking ensures this with an extensive portfolio, recently supplemented by mobile core technology from the 2023 Athonet acquisition. The HPE Aruba Networking Edge Services Platform (ESP) offers a secure, reliable, and AI-assisted network environment from the edge to the cloud. This way, businesses are always prepared for any present or upcoming connectivity requirements, be it via wired network, wi-fi, SD-WAN, or 4G/5G mobile communications.

With rock-solid connectivity, they are equally well-equipped for a digital twin of the factory floor, AI or robotics projects, the developing data economy, and sustainability initiatives – and for sending data from their servers to the focal point where manufacturing trends converge: reliable, low-latency, AI-optimized production.