Following exciting announcements at SAP’s Sapphire conferences earlier this year, I’ve had multiple thought-provoking conversations with clients eager to take advantage of the new technologies that PwC and SAP can offer. As I reflect on a busy first half of 2026, here are my top ten takeaways that remain front of mind as we head into the second half of the year.
#1 The conversation around AI has matured
This year, it feels like we’ve reached the moment of truth when it comes to AI. There’s still a big push towards AI. However, rather than boasting about the number of bots they’ve deployed, companies are prioritising quality over quantity.
Meanwhile, some of the topics discussed last year, such as providing structured data and finding ways to maximise AI efficiency, have now become an operational reality.
#2 SAP has set out its vision for the role of ERP in the coming years, under the banner of Autonomous Enterprise
SAP’s headline announcement at Sapphire was the Autonomous Enterprise: a bold vision where AI and humans work in harmony to drive real business outcomes. Internal core processes run automatically, and the human’s role is to steer and optimise.
This vision addresses a myth currently circulating in the market: that ERP will no longer be required, because agents will handle the work in the future. In the Autonomous Enterprise, SAP shows a different set-up grounded in practical realities: ERP will continue to play a central role in how businesses operate, because companies need a structural platform that is compliant and can run transactions at scale. Agents on top create content and drive efficiency, but structured data, compliance and security are all rooted in the ERP. As stressed in the conversation between Jeremy Barnum, CFO of JPMorganChase, and Christian Klein, CEO of SAP, on stage in Orlando, financial institutions managing trillions of dollars in payments every day cannot tolerate 80% accuracy.
I understand and agree with this focus on compliance, because implementing any new technology comes with risks. If a client automates invoices using AI and they issue an incorrect invoice, they’ll still be held accountable – they can’t simply blame their AI provider. This focus on accuracy, compliance and assurance fits well with PwC’s organisational culture. From our auditing business, we know exactly what companies need to be compliant, where to look, how to check, how to prove and how to approve. We believe these capabilities will become even more important as clients seek to deploy AI across their businesses.
#3 We’ve passed the peak of the AI hype curve, and companies are asking questions about cost
Companies are increasingly concerned about the cost of AI. Several people I’ve spoken to recently mentioned two new Gartner studies on this topic. One shows that spending on AI agents is rising exponentially – up from $86.4 billion in 2025, and forecasted to hit $206.5 billion in 2026 and $376.3 billion in 2027. Furthermore, a second Gartner study shows that by 2030, cost per resolution for generative AI (GenAI) will exceed $3, higher than many B2C offshore human agents. This shows that companies need to carefully consider their options for each use case, rather than assuming that AI is always the best route.
In cases where the business case for AI stacks up, it’s important to tailor the solution to the specific client situation, rather than treating AI as a one-size-fits-all tool. Our recognition of this and our willingness to meet clients where they are were among the key factors that contributed to us being named a Leader in The Forrester Wave AI Consulting Services report in Q2 2026.
#4 Joule Studio enables you to develop – without being a developer
One of the Sapphire announcements that stood out to me most was the launch of Joule Studio as part of the new SAP Business AI Platform. Joule Studio creates a fully managed environment for building and managing the full lifecycle of AI agents, applications and workflows, natively grounded in SAP business data, processes and semantics. It gives clients a quick, convenient way to optimise their SAP applications to meet specific requirements – with no knowledge of coding required. The idea is that companies can move from idea to production in minutes, not days.
I'm proud to share that PwC was named a winner in SAP's Agent race to Sapphire, which focuses on building AI agents at speed in Joule Studio. This win is a brilliant testament to the talent and ambition of our teams, and clear evidence that we're not just keeping pace with this shift – we're helping shape it. We’re already working with Joule Studio, and we’re excited to unlock the value it can bring to our own business and our clients. We’re looking to offer services to help clients use Joule Studio more efficiently.
#5 SAP ECC customers committed to migration are being given access to select AI capabilities
In another big change earlier this year, SAP announced that SAP ECC customers committed to migrating to SAP Cloud ERP solutions will be able to access select AI capabilities on legacy systems until 2030. This gives clients a bridge rather than a cliff to climb. It’s a big relief for clients still building their cloud business case.
#6 We won an award for our internal AI-assisted SAP implementation
Six months into the year, we already have two big awards under our belts. Alongside winning the Agent race to Sapphire competition mentioned earlier, we were delighted to be named the Cloud ERP Champion at the SAP Innovation Awards 2026. This award recognises our truly transformative SAP Cloud ERP programme – replacing highly customised legacy SAP ECC systems and a fragmented mix of SAP and third-party software with a clean core SAP Cloud ERP platform, leveraging the breadth of SAP Business Suite and SAP BTP to create a unified, intelligent environment across 19 countries and counting. A huge thank you to the incredible PwC team behind this success.
This is now our 4th consecutive SAP Innovation Award win, as we’ve picked up awards in 2023, 2024, 2025 and now 2026. I believe this is a record-breaking streak unmatched by any other SAP Global Strategic Service Partner (GSSP). I couldn't be prouder of what this team continues to achieve.
#7 We’re excited about new partnerships focused on AI-assisted SAP implementations
In the first half of 2026, we’ve been furthering our conversations and engagement with companies working to streamline and accelerate SAP implementations using AI. For example, we are already running proofs of concept with Palantir, Conduct AI and Nova Intelligence. PwC Germany has also signed a partnership agreement with Conduct AI, including a joint business relationship. This is a strong signal of where we're investing as a firm and a capability we'll be looking to scale globally.
#8 SAP has announced a unified API policy
Earlier this year, SAP published a unified API policy, ensuring that only approved architectures and solutions are able to access SAP-generated data at scale. We understand the compliance reasons behind SAP’s decision to make this move, and we’re prepared to help clients determine what this new policy means for them.
#9 SAP is addressing client concerns about sovereignty
This year, SAP has increased its focus on sovereignty, with clients in Europe the first to benefit. SAP and Mistral AI have formed a fully sovereign, enterprise-grade AI stack for clients in the European Union, and Mistral Plus has now reached general availability on SAP’s sovereign cloud. Additionally, Cohere’s agentic AI platform, North, is also set to reach general availability in June 2026. This is good news for clients in the EU, who often face obstacles to deploying AI due to concerns over data residency and model origin.
We know that there is strong appetite for better sovereign AI options among clients in the Middle East, some Asia-Pacific countries, Canada and the US public sector. We look forward to seeing further sovereign options for these clients in the future.
#10 Increasingly, competitive differentiation will come from deciding where to use AI, and where to intentionally avoid it
I recently had an interesting discussion with Dominik Asam, the CFO of SAP. He found a very eloquent way of expressing something that we’ve known for a while: large language models are very good at creating content based on structured data. The better the data foundation, the better and more accurate that content will be. However, LLMs aren’t good at running transactions at scale. If you try asking an LLM the same question twice, you’ll often get two different answers. So if you use AI to manage transactions at scale, there’s a risk that two similar transactions will be treated differently, and one will be incorrect.
This conversation underscored two key messages for me. One is that companies need to be careful and intentional about where they deploy AI, because businesses will still require transactions at scale in the next ten years. The other is that ERP will continue to play a key role in providing high-quality data that AI can use to create better content.
Closing thought
The first six months of 2026 have given us plenty of ambitious announcements and interesting food for thought. We look forward to working with clients to capitalise on the opportunities these create, and answer any questions they have.
If you’d like to discuss any of the topics raised in this blog, or to share your own thoughts on SAP’s direction of travel, please get in touch. I look forward to hearing from you.
