Table of Contents
Few people understand SAP’s research into emerging technologies like blockchain, AI, quantum computing and the metaverse better than Martin Heinig and Yaad Oren.
Heinig heads New Ventures and Technologies, a group of several hundred people working in labs to define SAP innovation and long-term strategy. “We look at technologies that have the potential to disrupt the market,” he said.
Oren heads a subgroup, the SAP Innovation Center Network, which he likened to a high school for research projects. “Once they graduate, they move into the real world,” he said.
Heinig’s is one group within an SAP R&D function that he said is divided into three parts: academic groups like the Hasso Plattner Institute, namesake of SAP’s co-founder, which takes the longest view, and product engineering, which operates on a roughly two-year timeframe.
“We sit right in the middle,” Heinig said. “We’re looking at opportunities for SAP that are five-plus years out and then try to figure out what something can be. We create prototypes to find out if we can really build it and if it is feasible for customers.”
A common theme of these technological investigations emerged during separate interviews with Heinig and Oren at this month’s SAP Sapphire 2022 conference in Orlando. Much of the work serves the ambitious goal of extending business processes beyond the walls of an organization. Doing so requires breaking business processes into smaller pieces that can be securely shared between software systems and corporate entities. Besides being shareable, these new processes are designed to be autonomous, “smart” and composable so they can be endlessly reconfigured to do exactly what people need them to do.
The interviews are combined and edited for brevity and clarity. Heinig and Oren both were emphatic that mentioning a technology does not mean SAP is committed to productizing it.
What important technologies are you working on that could lead to major changes in ERP?
Martin Heinig: Quantum computing, definitely, but it can also be simulated quantum computers like digital annealers, where we have a step change in computing power that can open up new scenarios. For example, in the supply chain, when you have optimization problems that would not take hours but minutes.
Things like homomorphic encryption can also be a game changer. The beauty of it is you can do analytics on encrypted data, so it will not reveal the actual information but you can still do some basic calculations. For example, I can give you sales data, but it would be encrypted so you don’t know the company that I’m working with, but you would see the order number or quantity. Whole industries could package the data and do analytics. It could be interesting in the healthcare sector, where you don’t want to reveal patient names.
The problem is it still requires a lot of computing time, so we need to go into the hardware space with partners and find out if there are some technologies, like specific chips, that can be a kind of coprocessor to minimize the penalty on the computing side.
When do you think quantum computing could be practical for business use?
Heinig: We see a lot of progress, and the number of qubits is increasing tremendously, but we have not found a quantum computer that can solve real-world problems yet. It’s hard to estimate, but it’s not 10 years out. Maybe the first real use cases are three to five years out.
We’re currently testing how it would work. The scenarios we are looking at are more in the optimization space, like supply chain warehouse management or production planning, where you have a lot of very complex problems to solve that need a lot of compute power. We try out how to translate these kinds of problems into quantum computing language.
Why should people care? What’s going to be so great about quantum computing?
Yaad Oren: It’s a whole new paradigm for computing. It’s not only the computation but how people will build software.
With classical programming, you interact with the processor in a certain way. If you have a quantum processor, you interact differently. Even the development languages get affected.
The disruptive potential is across the stack, from infrastructure to platform to software development languages.
There is a lot of hype, but SAP is currently looking at three areas where we see the potential for disruption.
The first is optimization. Quantum computing is not good for arithmetic, like one plus one equals two, but it’s very good for combinatorial problems like the traveling salesman problem, when you have many nodes and a factorial level of complexity.
Optimization problems fit quantum computing like a glove — for example, supply chain optimization, when you have so many parameters to evaluate regarding the route, pandemic regulations and weather.
We’re using a lot of quantum simulations — partner solutions, quantum annealing and other technologies — because the quantum computer is not there yet. SAP is also involved in a government-funded project with another German company on building quantum as a service. We use a lot of simulation technologies that have already helped us understand the power of this.
The second is called post-quantum cryptography, which is the security and encryption aspect of quantum, a big thing given the number of phishing attacks, ransomware and password hacks.
Quantum holds a lot of promise to create encryption at a level the industry has never known before. It’s about creating a new type of password that is not breakable. The quantum code is the means to the end.
In browsers today, you have auto-generated passwords that are done by algorithms. Quantum computing can give birth to new types of algorithms that create passwords at a new level of complexity.
Eventually any password is breakable if you spend enough time. With quantum computing, this becomes much harder or impossible. Of course, it’s a matter of time until hackers catch up.
The third benefit of quantum computing is AI. Machine learning is based on unique data, and you need computation power to train models. With quantum computing, you can create new types of AI models and applications that you couldn’t train before, because now you have a strong computer that can learn more and solve new problems. It will give birth to new types of automation and predictive analytics.
AI and machine learning
What kind of research are you doing in AI and machine learning?
Heinig: Enterprise knowledge graphs are a concept based on machine learning technology that we’re looking into. It’s basically the idea of modeling the connections between business objects and bringing in the relationships between them. This is a very important technology for creating context for situational awareness and personalization.
Oren: Regarding the future of AI and analytics, we have a lot of advances in this area. SAP is also focusing on infusing AI into the core application.
We are working a lot on the future of planning and introducing new types of AI like reinforcement learning to create new types of simulations.
Today, if you want to have planning solutions, you cannot always get the full perspective on uncertainties, and you cannot have recommendations and simulations for scenarios that you didn’t ask about.
We’re working on a self-learning system that provides continuous intelligence. It’s not a product yet, but we’re working on it with customers. You don’t need to train the model and build the machine learning model yourself. It can keep learning even in areas that you didn’t explicitly ask to explore, to fight uncertainty. This was requested by customers during the coronavirus pandemic and all the disruption in supply chain, where you need to deal with a lot of uncertainty.
Does the metaverse have implications for ERP, realistically?
Heinig: Yes, but the question is when and to what extent? What does it mean from a process perspective? Companies have already crossed the borders between physical goods and digital goods. The basic research question that we are looking into is how can we make these processes seamless?
You have your physical store where you sell physical goods and you have a digital store. You can sell physical goods, order them and get them delivered. So how can we extend this so you can also sell in your digital store a physical good with a digital good, like a non-fungible token (NFT)? No matter where you want to do business in the metaverse, the ERP system should help you run your processes.
Where do you see the most promising business applications of the metaverse?
Oren: The metaverse is also a lot of hype, and we need to distill the noise. For me, the magic happens more in the practical — I would even say boring — side of the metaverse, not the avatars and the UIs.
We’re looking more at the Web 3.0 aspects of things. Web. 3.0 is kind of what greases the skids of the metaverse — for example, all the crypto payments.
Of the top three long-term observations we have about the metaverse, the first one is everything regarding digital finance. There is huge demand from the industry. The number of transactions and volume of the new generation of buyers is huge — how you accept crypto payments and how you sell NFTs.
Let’s say an avatar is buying something. You need a profit and loss statement that can take fiat money and crypto money. How you do the balance sheets and audit them may not be sexy, but it’s really important.
The metaverse is a combination of real-world technologies and the digital world. How do you do analytics and planning if you have functionalities and workflows and things that are both digital and real? Those are different areas that need a bridge between them.
The third thing is the augmented employee. They’re going to have digital representations in the metaverse. We’re evaluating how you can connect those representations into the enterprise system from SAP SuccessFactors to all the other data sources you have in the organization.
Some members of the U.S. Congress asked the Environmental Protection Agency to consider regulating bitcoin miners because they use a lot of compute power and water. Are you trying to improve the efficiency of blockchain?
Heinig: We take this into consideration, especially proof-of-work authentication, which is very energy consuming, but it’s not what we research. It’s more about how would we use blockchain technology, hopefully in a very energy-efficient way, to find customer use cases we can enable with blockchain.
One good example would be a green token for tracing raw materials, using the concept of tokenization and blockchain technologies in cross-company scenarios.
Another idea is cross-company workflows. For example, how you can have different process steps across companies and across systems, store them and make sure they are auditable.
A third example would be self-sovereign identity (SSI). The idea is you store your identity in a personal wallet. Today you have a central register where you store the identities and proof of identity against one central database. A good example is when you use your Google identity to log in at different websites.
The idea is similar to a bitcoin wallet. You would have credentials that are verifiable in your personal wallet, and you can verify yourself against different systems. We would make sure it is auditable so you always know that an identity is real.
The beauty of this concept is that you can work with different systems seamlessly. Maybe in the future it would be a way to have more personalized experiences with systems because it could also store information that a system could use to personalize your experience.
What blockchain mechanisms are you looking at for connecting business processes and building trust?
Oren: It started with onboarding. Let’s take a supply chain or order-to-cash — any process with many vendors. Today, when you onboard a new player to a business network, there are a lot of time-consuming manual steps and authorization. The mechanism we use, self-sovereign identity, harnesses the power of a blockchain so that everything is auditable and immutable. You can quickly onboard vendors to the network, supply chain and any process.
With blockchain tokens, you can onboard vendors with ease because everything is documented. In enterprise processes like order-to-cash, any step, like when you deliver something — let’s say you’re manufacturing an engine, to use an example from the keynote — you deliver the piston, someone else provides another component. For anything you send between vendors, you need to have proof-of-delivery documents, which are legal documents. You need to call a lawyer and have a notary service sign the document. It is paper based.
Using the token, you can do self-authorization. You don’t need to call those legal services to sign documents. Using the blockchain, everything is immediate, auditable and transparent. It’s part of a proof of concept. It’s not a product but they talk about it publicly.
We also have this carbon data network project that was also mentioned in the keynote where you have track and trace to see the CO2 emissions of each part by each vendor in each stage of the supply chain.
You are doing some research on composable business processes. What specifically are you looking at?
Heinig: We have 50 years of business knowledge that’s basically all packaged in our S/4HANA system. How do we find a way to make it composable to make it more flexible and include easier third-party solutions?
Integration today is basically happening on a technical level, but we would like to lift it up on the business process level. Today we sell software that’s packaged, and you have the business processes inside the software. But I think we should change that so we would sell you business processes and you would not even need to bother with what kind of software you’re using, because these would be packaged, orchestrated functions that are already pre-integrated.
Oren: If you have, let’s say, a need in order-to-cash for a new type of vendor verification or some compliance, it should be very easy — like plug and play — to add services from SAP or not. We want to have this orchestration layer of having two services working together. This is something that requires a lot of technical underpinning to both have the abstraction and orchestration of services to work together.
Analytics is a major focus of your group. Why is analytics worth looking at?
Heinig: It’s basically analytics plus planning, and we see two major differences arising.
One is the role of ERP systems and business networks. Let’s take sustainability KPIs. It’s not enough to try to analyze and optimize them on a company level. You need to look at the whole supply chain on your business network. This means your analytics capabilities need to go cross-company.
From a planning perspective, if you really want to optimize it, you also need to have these planning capabilities along your whole supply chain. This is where things get really, really complicated.
The second one is around how can we lift up analytics and planning to the next level? Today, it’s really manual and static. You look at your dashboards and maybe find some anomalies and try to react.
We’re trying to change this so it’s possible for the system to automatically detect anomalies in data flows and trigger creation of a dashboard that is personalized to your role in the company. The system says, ‘we found something, please have a look at it, and these are your three most appropriate options.’