How Nestlé is putting the ingredients in place for embedded AI with three-stage S/4HANA upgrade

Long-term SAP user Nestlé has completed the first stage of a major project switching from ECC to SAP S/4HANA Cloud Private Edition. The first part of a three-stage project, the current rollout covers 112 countries and more than 50,000 employees.

The food and beverage giant owns over 2,000 brands across the world, including Nescafé, Perrier, Carnation, and Christmas favorite Quality Street. It has been using SAP as a single unified system since 2000, and shifted to the cloud in 2022.

A big impetus for the switch to S/4HANA was the fast-approaching end of mainstream support for ECC, which runs out in 2027. Nestlé was also keen to take advantage of the single template to gain a single quality source of master data. Ralf Huebenthal, Global Head of IT Platforms at Nestlé, explains:

If we’d have stuck with ECC, we obviously would be missing out on all these opportunities in terms of integrating AI in our business process, simplification of business processes and so on. That's why we are doing it now.

Green, brown or blue?

The first decision Nestlé had to make was, which kind of upgrade to do: greenfield (building a new SAP system from scratch); brownfield (upgrading the existing SAP system while retaining the current configuration and data); or bluefield (a hybrid approach combining elements of both so firms can use new SAP features while retaining existing functionality and data). Huebenthal says:

For us, it was relatively quickly clear. We have to do brownfield as we have a single template, as we have well-established shared service centres, and as we are well standardized in this area, so we can drive or can operate with scale and speed.

The project has been running for around 1.5 years. The actual update to S4/HANA took 13 days in total, with just 19.5 hours of downtime, which Nestlé managed to achieve via a lot of testing and rehearsing.

Because of the single development track undertaken by Nestlé, systems are standardized across the organization, and the firm then has three global instances. It was already operating in the cloud before the S4/HANA rollout. Huebenthal notes:

We moved to the cloud first because then when I do all the testing in the cloud, it's a whole lot easier for me to get a 24 terabyte (TB) instance, which I can spin up in days, not in months if you do it in your own data centre, considering all the cabling and so on you have to do. That was one of the key drivers why we went to the cloud, and we have seen that now during our S/4 upgrade, we were able to have all the test instances and we could spin them up and remove them as we needed, which gave us the flexibility and agility for testing.

When Nestlé started the project, Huebenthal says it was the first company using 24TB nodes fully virtualized. At that point, Microsoft was the only provider able to give the company 24TB fully virtualized and procured through SAP, which he cites as the reason for selecting the private cloud edition. 

The first rollout with the 24TB node covers 112 countries operating across Asia, Oceania and Africa. The next stage in the project is Europe, which Huebenthal says will be another 24TB node, potentially going to 36TB; the last stage is North and South America, which has a scale-out 48TB node. He adds

We have to do it step by step obviously to mitigate the risk to the business and to allow enough testing time and also to allow enough time to plan the business downtime. While it's only 20 hours, compared to what I've heard from many other companies, it's still 20 hours’ business downtime and the windows we can have business downtimes are very limited. We have to find always the right seasons, the right timing so that we don't impact the business in any form negatively.

Embedded AI

While Nestlé’s initial move to the cloud was predominantly a lift and shift, the move to S4/HANA future-proofs the business and gives it access to new capabilities SAP is releasing on an ongoing basis. This is particularly around Joule and the ability to embed AI in its processes, where the company already has some proof of concepts and pilots happening. Huebenthal adds:

That's really the key capabilities we are looking forward to get out of this S/4 upgrade.

Nestlé also took advantage of cleaning up its core as part of the project. While it’s not at 100%, as its core has got a lot cleaner, the firm can upgrade industry capabilities much faster. The S/4 upgrade has also increased its ability to more accurately plan when it delivers, what it delivers and how it produces.

While the decision to stick with SAP comes partly down to the two businesses having a solid partnership after 25 years working together, there’s also a more realistic reason. Huebenthal says:

We are running on SAP globally, more than 80% of our turnover runs through SAP. If you would move away from that, just the cost of doing so you need to get a lot of value out of the alternatives which are there on the market.

Customers are already benefiting from the rollout, for example via Advanced Available-to-Promise (aATP), which provides a response to order fulfilment requests and is enabling Nestlé to do more accurate planning over which products it delivers to which customers. Joule and AI will simplify a lot on the finance side when it comes to closing repetitive and redundant tasks that AI can take over from people. Huebenthal says:

For me, sitting on the technical side of things, it's quite speculative to talk on behalf of my finance or supply chain colleagues. But obviously the ability for a user to integrate or to work with natural language through AI will unlock a lot of benefits and speed we don't see today.

Nestlé is already quite advanced when it comes to AI, and currently has more than 100,000 users using the technology everyday through Copilot or its own NesGPT.

Network upgrade

As well as the SAP upgrade, late last year Nestlé began a complete network transformation, which will bring more digitalization to its over 300 factories around the world. Huebenthal explains:

We need to upgrade our LAN and Wi-Fi in these areas. And we are moving to a Wi-Fi first approach, which helps us to provide product coverage for aspects of our production lines in the factories, allowing users or production line workers on the line to use iPads, iPhones or other Android devices, whatever is the right device, and to track more accurately what's happening in our production lines.

The company is also undertaking a huge network transformation of its Wide Area Network, moving its network backbone from its current provider to Google, which has increased speed by 40%. Huebenthal notes:

That speed increase is noticeable to the end user. In particular, if you talk about AI, you need an AI-ready network.