‘The AI hype has value’: SAP’s Muhammad Alam speaks to the ROI debate
Businesses can often find themselves caught between two competing narratives. On one side, there's the relentless hype suggesting AI will revolutionise everything overnight. On the other, there's growing skepticism as companies struggle to translate AI investments into measurable returns.
Accentuating this gap is recent research. A July 2025 MIT study found 95% of enterprise AI deployments fail to deliver value. Meanwhile, in a report published October 2025 by The Wharton School found that 75% of business leaders report a positive return on investment (ROI) from their AI investments. Fewer than 5% say returns have been negative.
As Muhammad Alam, Member of the Executive Board of SAP SE, Product & Engineering, attested to, the truth is somewhere in the middle and there are specific ways companies can enhance the value they're seeing from AI initiatives.
When AI doesn't exist in a silo, it creates more value
The architecture prior to AI implementation matters a great deal. According to Alam, it's crucial to understand that AI is not a standalone solution but as an integrated layer within existing technology stacks.
He says, "Fundamentally, for the first time in decades, there's a new layer that's being added to the tech stack: the AI layer."
He continues: "AI can't exist without the stack underneath. You can't have business applications without a platform or an infrastructure underneath. The AI layer has new needs, and not necessarily just with the data in the app, but the layer underneath. Where else does the data get created? If you don't have apps, where else does the action get taken? From the AI, from agents? If it's not in the applications you need the level of governance, the structures of localisation, the compliance for the applications."
According to Alam, if AI is implemented as a separate entity sitting on top of existing systems, it inherits all the complexity of finding and accessing data. Ensuring the right authentication, security, governance, and more, adds exponentially to implementation tasks, and that doesn't even account for where the final outcome will be available for end users. This is a huge driver of value.
He says, "When AI comes seamlessly together with the application layer that produces the data it needs to run, it creates the most value."
Breaking down data silos to realise ROI
Alam emphasises that data fragmentation is one of the biggest barriers to AI ROI. The pattern of failure is consistent: companies bolt AI onto legacy systems, creating new integration challenges rather than solving existing ones. Data must be extracted, cleaned, harmonised, secured and then fed to AI systems. Otherwise, each step can add cost and potential points of failure.
Alam comments, "It's not that we believe that the tech leaders don't want to [innovate with AI], but the way they're approaching it makes it a lot harder and a lot more complicated."
"Where we see success is where people are taking a more holistic approach. For instance, for our application landscape, we've got the app, data and AI coming together - let's enable them, because the data is already part of the application. You pass the same sort of construct to the AI. The AI is embedded in the application, and then you can deliver value and ROI to the organisation. And if you need to extend or build custom, then you can use the same framework to then extend it to other parts of your organisation, too."
SAP's embedded AI strategy
To enable this level of seamlessness, Alam says SAP has focused on creating AI that is embedded into applications, offering the potential to create seamless value and ROI that would otherwise be difficult and cumbersome for customers to achieve on their own.
SAP's strategy leverages its unique position in the enterprise software landscape. The company provides applications that cover a significant breadth of business processes - finance, spend management, supply chain, human capital management (HCM) - ensuring data is harmonised and seamlessly embedded in applications through Business Data Cloud.
Alam says, "We believe this ability to create unique use cases that seamlessly bring app, data and AI together - alongside positioning it, first and foremost, as there to help the individual, the human, get smarter, more efficient, more productive, leading up to autonomous execution and creating significantly more value - is the strategy that will resonate as bringing value to end users."
It's a fundamentally different approach than the 'AI-first' mentality that encourages companies to experiment with AI before understanding where it fits into their operations. Instead, SAP is betting that AI delivers the most value when it's invisible, when it enhances existing workflows rather than creating new ones.
Adaptability and looking ahead: Start with understanding, not technology
The AI hype presents both opportunities and challenges. On one hand, executive enthusiasm has unlocked budgets and created organisational momentum. On the other, the pressure to implement AI quickly can lead to poorly planned initiatives that fail to deliver lasting value.
Alam comments, "The hype is creating the belief that real value can be achieved with AI. And not just that it can be achieved, but it needs to be achieved now and as quickly as possible, because others are doing it, and if not, we will get left behind. With that comes budgets, commitments and organisations that are there to help make that happen, which I think is a very positive development."
But Alam quickly adds a crucial caveat: "However, this is where I go back and say: now that you have your AI application or implementation, how do you make sense of it? Just because you can generate an app through a large language model (LLM), as some of the LLM providers would say, does it make sense for you to go generate a procurement app? What are you going to do next year when you have to enhance it, when you have to have compliance around it, when you have to make sure it's in different languages?"
Trust and adaptability must be considered from the start for AI to deliver lasting value.
As highlighted by SAP's executives, first, understand your company and identify the specific ways in which you could be faster, more efficient and more productive. This is where you can begin to define roles and build a step-by-step deployment of autonomous execution. The maturity of the value journey will follow naturally from this foundation. Critically, it's much more difficult to identify appropriate roles and build effective applications once AI has already been deployed. The technology-first approach often leads to solutions searching for problems - expensive experiments that never scale beyond pilot projects.
In addition, Alam encourages businesses to start with functions that more easily lend themselves to AI and LLMs. Human resources and customer service, for instance, offer clear use cases with measurable outcomes and relatively structured data. These areas provide opportunities to build confidence, demonstrate ROI and develop organisational capabilities before tackling more complex implementations.
At the end of the day, end user experiences are everything
For all the discussion about infrastructure, data architecture and AI models, creating value comes down to creating powerful end user experiences. Infrastructure must serve experiences, not simply exist for its own sake.
As Philipp Herzig, SAP Chief Technology Officer and Chief AI Officer, explains, the disconnect between AI potential and current results stems partly from unrealistic expectations about timelines and impact, and the lack of emphasis on user outcomes.
Herzig draws a parallel between AI adoption and the evolution of the internet, offering a roadmap for where the real value will ultimately accrue.
He says, "To me AI is comparable to the internet. In the beginning of the internet there was a lot of focus on the infrastructure, for instance, the companies that built the switches and the routers. Then the discussion moved to the ISPs, the AOLs and telecoms of this world, that brought the internet to the home and built the distribution to all the consumers. Eventually, the value accrued in cloud services that we already have, streaming services, cloud software, SAP software, and so on."
As Herzig shares, while initial investments and attention focused on enabling technology, the infrastructure layer that made everything else possible, the lasting value and market dominance eventually flowed to companies that delivered superior end-user experiences built on that infrastructure.
Herzig continues, "When it comes to AI, my belief is that even though there's a lot of discussion currently on the hardware side, data centres and GPU capacity, this is not where the end game will be played. The end game will be played with amazing end user experiences, AI-native software-as-a-service that people can just turn on. We will see many new business models that also will emerge out of this. In addition, different ways of consuming these AI experiences, not just from a user experience but also from a commercial experience perspective, is what we are converging towards in the next couple of years."
The path forward: Integration, not isolation
Ultimately, companies seeing genuine ROI from AI are those that have integrated it thoughtfully into existing workflows, rather than bolting it on as a separate layer. They've broken down data silos, embedded AI capabilities directly into applications, and focused relentlessly on end-user experiences rather than technology for its own sake.
As the hype cycle matures and AI moves from experimental to essential, SAP's executives highlight that the winners will be those who understood early that great AI needs great architecture - and that architecture must serve human experiences, not the other way around.
