Geneva Insights September 2025

September 30, 2025

Executive Summary

September 2025 brought a concentrated series of AI events to Geneva, revealing a fundamental tension in AI adoption for wealth management: the choice between technological sovereignty and cutting-edge capabilities. While Switzerland’s Apertus continues developing its open approach, the month’s events and discussions illustrated the practical challenges institutions face when evaluating AI solutions.

Germany’s announcement of a partnership with OpenAI rather than European alternatives also came up in conversations, raising questions about how organizations balance capability with sovereignty concerns. Throughout the Swiss AI Weeks, fintech professionals, cloud providers, and researchers exchanged perspectives on the same underlying question: How do we balance regulatory compliance with competitive advantage?

For wealth managers, September’s developments revealed that infrastructure decisions – where your AI runs, not just which model you use – have become strategic choices with lasting implications.

September in Geneva: AI Weeks Bring Theory to Practice

September transformed Geneva into a hub for AI discussions, with the Swiss AI Weeks hosting multiple events focused on practical implementation in financial services and beyond.

Apertus: The Swiss AI Alternative

The month opened with continued developments around Apertus, Switzerland’s first large language model that we covered in detail in our Special Edition. Launched through collaboration between ETH Zurich, EPFL, and the national supercomputing center CSCS in Lugano, Apertus represents Switzerland’s commitment to AI sovereignty through its “fully open” approach – publishing training data, code, and intermediate checkpoints.

While Apertus still needs to catch up with the capabilities of the latest US and Chinese frontier models, the possibilities for Swiss actors to develop tools in Switzerland for Swiss and European clients remain an exciting development. The critical question for wealth managers is not whether Apertus matches the most advanced models today, but whether having a European alternative with full transparency and local deployment options provides strategic value.

The Geneva FinTech Gathering: Practical AI Challenges

The Geneva FinTech and Sustainable Finance Community hosted a meeting during Swiss AI Weeks on “Shaping the Future of Finance,” bringing together local actors from Geneva’s tech and financial communities to exchange ideas.

The discussions focused on practical AI use cases among Swiss financial institutions. What became clear throughout the conversations: navigating between technical possibilities and regulatory hurdles has emerged as one of the crucial points to consider for successful AI implementation.

AI for Society Conference: Broader Perspectives

Late September brought the AI for Society, Geneva Generative AI Conference at Campus Biotech. The conference provided a broader overview across industries, bringing together researchers, practitioners, and policymakers.

Among the various topics discussed, one strategic question came up repeatedly: the recent announcement that the German government would partner with OpenAI rather than European alternatives for its sovereign AI infrastructure. The decision illustrated the practical trade-offs that organizations across sectors face when implementing AI – balancing capability requirements with sovereignty concerns.

The German Decision: A Strategic Choice

On September 24th, the German government announced a partnership with SAP and OpenAI to launch “OpenAI for Germany.” The sovereign AI solution will establish German data centers and create a specialized version designed for the German market, keeping government and enterprise data within German borders while providing access to OpenAI’s models.

The announcement raised an important strategic question: Why didn’t the German government consider European partners such as French Mistral or Swiss Apertus, but chose a US provider instead?

While geopolitical considerations are beyond this newsletter’s scope, the regulatory implications merit examination: when regulators impose strict data protection requirements on companies concerning sensitive data, yet governments themselves – entrusted with the most sensitive data about their citizens – partner with US providers, what practical framework does this suggest for balancing capability with sovereignty?

Germany’s decision suggests a framework where data sovereignty can be achieved through infrastructure and contractual arrangements rather than requiring European ownership of the underlying technology. For wealth managers making similar decisions, this raises important questions about regulatory expectations: While Germany’s approach suggests that data residency and operational controls may satisfy sovereignty requirements, it remains to be seen how local regulators like FINMA or BaFin will evaluate such arrangements in the financial sector. Institutions should consult with their legal and compliance advisors regarding specific regulatory interpretations and requirements. Can institutions reference the German government’s decision as precedent, or will financial sector regulators apply different standards?

Infrastructure Events: The Foundation Layer

Cloud providers also hosted events in Geneva’s Silicon Chalet Series during September. Swiss provider Infomaniak and Oracle each presented on AI infrastructure options, providing overviews of the current state of the art in AI hosting.

The options range from helping companies with on-premise hardware to providing locally hosted AI “as a service” to full cloud solutions. Depending on what institutions need in terms of data protection and regulation, various options are becoming available to support wealth managers’ AI adoption.

These infrastructure discussions matter because where your AI runs, where data is processed, and where the output is used determines what regulatory framework applies. A wealth manager using cloud AI from a US provider faces different compliance obligations than one using locally-deployed European models, even if both provide similar functionality. Additionally, the location of your clients matters: serving EU clients triggers EU AI Act obligations regardless of where your firm is based.

The infrastructure discussions reinforced a crucial point: where AI systems run – on-premises, in Swiss data centers, in EU cloud regions, or in global infrastructure – shapes everything downstream, including which models become available, which regulatory requirements apply, and which vendors become possible partners.

The Sovereignty Question: What September Revealed

September’s events in Geneva illuminated a tension that wealth managers must now address directly:

Regulatory frameworks demand data sovereignty and transparency. Competitive pressures demand cutting-edge capabilities. European alternatives that maximize sovereignty often trail in capabilities. Frontier models that maximize capabilities typically come from outside Europe.

Germany’s OpenAI decision suggested one approach: prioritize capability while achieving sovereignty through infrastructure rather than provider nationality. This resolution raises questions:

Can contractual guarantees about data handling truly substitute for European ownership of AI infrastructure?
If a US company operates data centers in Switzerland with appropriate controls, is that sufficient?
How should institutions weigh geopolitical risks against capability advantages?
Does choosing European alternatives today, accepting current capability gaps, become strategic if those alternatives improve over time?

There is no universal answer. The right approach depends on each institution’s specific risk tolerance, regulatory exposure, client expectations, and competitive positioning.

Practical Next Steps

Based on September’s developments, wealth managers should consider three immediate actions:

1. Inventory Your Current AI Posture

Document where you stand:

Which AI tools are employees using?
Where is compute happening?
Which vendors have access to what data?
Are current deployments EU AI Act compliant?

2. Engage With Infrastructure Providers

Have conversations with Swiss cloud providers, global providers’ EU offerings, and on-premises solution providers. Understanding infrastructure options shapes possible AI strategies.

3. Evaluate European Alternatives

Conduct actual testing with European models on representative use cases. Strategic decisions require empirical data, not assumptions.

EU AI Act: A Practical Resource

As these strategic decisions become urgent, we are offering a resource to help navigate the regulatory landscape. Our recent webinar “EU AI Act in 10 Minutes” provides a concise overview of what wealth managers need to know about compliance obligations.

Access this and previous AI x Wealth Management webinars through our collection.

Looking Ahead

September’s events in Geneva transformed abstract debates about AI sovereignty into concrete decision requirements. The Germany-OpenAI partnership demonstrated one approach: prioritizing capability while seeking architectural solutions for sovereignty. The Swiss AI Weeks showed a maturing ecosystem of European alternatives that offer distinct advantages in transparency and control, even if they don’t yet match frontier capabilities.

For Geneva’s wealth management community, the path forward requires balancing regulatory compliance, competitive positioning, client expectations, and geopolitical considerations. There is no single “right” answer, but there is increasing urgency to develop a thoughtful approach.

The institutions that will thrive are those that approach AI adoption strategically – defining sovereignty requirements explicitly, evaluating infrastructure options systematically, and building optionality rather than vendor lock-in.

The EU AI Act’s obligations are approaching. Competitors are deploying AI at scale. Client expectations are rising. The window for deliberate strategy before being forced into reactive responses is narrowing.

Do not hesitate to contact us if you wish to discuss options with an independent strategy advisor that understands your specific business needs.

This is the fourth monthly edition of our AI x Wealth Management newsletter. Your feedback continues to shape our focus – which developments would most benefit your practice? Share your thoughts in the comments or connect with us directly via LinkedIn message.

Sources:

About the Author: Dr. Andreas K. Janoschek specializes in AI applications for Asset & Wealth Management. Based in Geneva, he helps industry professionals navigate the intersection of finance and technology.

This newsletter aims to inform and does not constitute investment, legal, or regulatory advice. Always consult with qualified professionals for specific circumstances.