Everyone in Silicon Valley is asking whether AI will kill SaaS, but Europe should be asking something else: where value is moving and how it can capture it.

The “death of SaaS” debate goes something like this: AI agents will generate software on demand. Instead of buying SaaS products, companies will spin up bespoke tools. The result? SaaS margins collapse, software multiples compress and a new AI layer captures most of the value.

It is an interesting debate, but it frames the opportunity too narrowly for Europe. The shift underway is not about SaaS disappearing, but about how it is used, where value accrues and what gets automated next. If you look closely, this transition may play far more to Europe’s strengths than people realise.

There is also a reason this debate is coming out of Silicon Valley. The US has built the most successful SaaS companies in the world over the past 15 years. Predictable revenue, high margins and leverage at scale made SaaS the ideal asset class for both venture and private equity. What AI is threatening is not software itself, but the predictability of those revenue streams and the margin structure behind them.

Let’s start with the obvious. SaaS is not going away. The reality is more nuanced.

Enterprise software like Salesforce, SAP or ServiceNow is not going to be replaced overnight by some AI-generated script. These systems are deeply embedded in business workflows and reflect decades of testing, compliance layers, integrations and institutional knowledge.

What is changing is where value sits. Historically, SaaS products were the interface. You opened Salesforce to manage sales, Notion to manage documents and Slack to communicate.

Now we are moving toward a different model. Instead of opening software, you ask an AI agent to complete a task such as writing a report, summarising a meeting, drafting a contract or analysing a pipeline. The agent pulls data from multiple systems and executes across them.

In that world, SaaS tools do not disappear. They become infrastructure, and value moves to the orchestration layer.

Underneath this, a more fundamental shift is taking place. For decades, software helped people perform tasks. Now it is starting to complete the work itself.

This also changes how value is measured and paid for. In SaaS, value is tied to usage. In services, value is tied to outcomes. AI collapses that distinction, moving pricing power away from per-seat models and toward value-based pricing as software gets closer to delivering outcomes directly.

Instead of paying €100 per user per month, companies may pay €50,000 to complete a legal analysis, design an engineering system or run a research simulation. At that point, software stops looking like SaaS and starts looking like a services business, except the services are delivered by software.

For large, mature companies, this creates pressure. For newer companies, it opens a different design space. Companies that fail to adapt their pricing and value capture models will struggle, while those that do will look less like traditional SaaS and more like outcome-driven platforms.

This is also where the “death of SaaS” narrative actually comes from. Not from new startups. From existing portfolios. When margins compress even slightly in highly valued SaaS businesses, the impact on equity value is disproportionate, especially for companies priced on predictable expansion and operating leverage.

The real disruption is not happening inside the software industry. It is happening inside the services economy. So part of what we are seeing is not just a technological change, but a repricing of an entire asset class.

From an allocator perspective, this creates an asymmetric setup. Part of the existing portfolio may come under pressure, but the upside from companies that capture this shift can more than offset that. Net, this is likely positive for venture portfolios, but unevenly distributed. It is more challenging for large, leveraged public SaaS companies than for early-stage investors building into the shift.

This is where Europe becomes structurally interesting. Not because of AI capability, but because of where value is already concentrated.

In this transition, Europe has an edge. Its economy is not built around consumer tech platforms, but around deeply embedded industries such as manufacturing, engineering, automotive, energy, pharma, financial services, consulting, compliance and industrial design. These are large, service-heavy sectors.

This suggests that the real opportunity for AI in Europe is not building another horizontal SaaS product, but automating entire service categories. Europe does not need to win the model layer to win economically. It needs to own the workflows where value is high, regulated and deeply embedded in real-world systems.

Imagine AI systems that design industrial components, automate regulatory compliance, run pharmaceutical research workflows or replace large parts of consulting analysis.

These are trillion-euro industries, and Europe already has a strong foothold in many of them. If software starts replacing services, Europe may be better positioned than people think.

Europe is unlikely to dominate every layer of the AI stack. Infrastructure is largely controlled by the US, where NVIDIA, hyperscalers and the capital required to build frontier compute clusters create a massive moat. Foundation models are also consolidating around a small number of players, including OpenAI, Anthropic, Google and DeepSeek. Europe will participate here, but it is unlikely to lead.

This concentration is already visible in model development. In 2024, the US produced 40 notable AI models, China 15, and Europe just three, creating advantages that are difficult to replicate.

That gap is driven by capital: US private AI investment reached $109 billion in 2024, compared to $9.3 billion in China and $4.5 billion in the UK. In generative AI, US investment exceeded the combined EU and UK total by over $25 billion.

The constraints are different, and once they change, where effort is allocated changes as well.

But the layers above this remain open: agents, orchestration, vertical applications and, most importantly, services automation. Here, expertise matters more than GPU clusters. And that is where Europe has real advantages.

For the last decade, Europe has spent a lot of time worrying about the absence of large consumer tech platforms. The next decade may look very different, as the shift is increasingly about where work gets done and how it is priced.

That is why the SaaS debate is a surface-level question for Europe. If software starts capturing outcomes rather than enabling them, value moves closer to the industries themselves, and that is where Europe already operates.

While Silicon Valley debates the future of SaaS, Europe faces a more relevant question: whether it can capture that value, and where it will concentrate, as software starts behaving like services.

That answer is still open.

For once, the next wave might not start in Silicon Valley. It might start in Stuttgart, Stockholm or Eindhoven.