Oracle Cuts 30,000 Jobs for AI Infrastructure: What It Means for Swiss SMEs

The Global AI Infrastructure Wave: Oracle, TCS and Microsoft Reorganise

Oracle is cutting up to 30,000 jobs across Oracle Health, Cloud and Consulting. This frees up $8 to 10 billion annually, money flowing directly into the Stargate data centre buildout, underwritten by the $300 billion contract with OpenAI. In parallel, other hyperscalers are building their own fronts: TCS is assembling a Forward-Deployed AI Engineering Unit with 5,900 to 8,900 engineers, and Microsoft is investing $2.5 billion in a Frontier Company Unit with 6,000 industry and engineering experts working directly on client projects.

The signal behind the number matters more than the number itself: hyperscalers are shifting their business model from licence sales to direct deployment of AI specialists at client sites. This intensifies global competition for exactly the specialists that Swiss companies also need for their own AI initiatives — talent is becoming scarcer even before most SMEs have reached implementation.

The Swiss SME Reality: Ambition Meets Novice Status

The Deloitte Switzerland AI ROI Report 2026 reveals a striking gap: 30% of Swiss companies plan to increase AI investment by 20 to 39%, well above the EU average of 23%. 52% expect a return on investment within one year. Yet 30% cite lack of technical talent as the top barrier, 27% struggle to identify viable use cases in the first place, and only 24% have mandated AI training for their workforce.

The OECD D4SME programme adds the structural picture: 34% of Swiss SMEs now use AI, up from 22% the previous year. But 76% of them remain at novice level, essentially using AI tools for email. Only 3.6% reach champion status. KOF ETH Zurich confirms this: 62% of SMEs limit AI use to basic tasks, and only 34% have defined any governance rules for its use.

Why the Infrastructure Expansion Reshapes the Build-vs-Buy Calculation

When hyperscalers such as Oracle, TCS and Microsoft deploy thousands of specialists directly into client projects, the cost logic shifts for anyone planning to build an internal AI department from scratch. Access to top-tier talent becomes scarcer and more expensive, while pure model costs continue to fall, lowering the technical entry barrier without solving the talent problem. This divergence is central: falling model costs make AI more accessible on paper, but without specialists who can identify use cases and operate systems reliably, cheaper models deliver little value. For most Swiss SMEs, the calculation is increasingly tilting toward specialised partners rather than internal buildouts.

Swiss AI Roadmap 2026: Structural Answers for the Mid-Market

The Swiss AI Roadmap, presented by digitalswitzerland in June 2026, addresses exactly this gap through seven strategic directions, from scaled AI education to world-class research, resilient infrastructure, and smart governance. Four concrete instruments becoming operational in 2026 are particularly relevant for SMEs.

• European Digital Innovation Hubs (EDIHs) providing low-threshold advisory and testing infrastructure for SMEs
• Innosuisse funding to co-finance AI pilot projects and capability building
• Regulatory sandboxes, operational since 2026, allowing low-risk experimentation
• Apertus, a sovereign Swiss language model for data-sensitive applications

The roadmap projects an annual innovation gain of CHF 15 billion by 2034, provided these structural offerings are actually taken up by SMEs and not only by large enterprises. The current gap between investment appetite and workforce readiness remains substantial across the Swiss economy.

What Swiss SMEs Should Do Now

1. Actively evaluate EDIH and Innosuisse offerings instead of waiting for internal capacity to materialise
2. Explicitly calculate build-vs-buy: compare the cost of an internal team against the total cost of ownership of an external partner
3. Define two or three concrete, measurable use cases instead of leaving AI use unstructured and confined to basic tasks
4. Establish governance rules early, rather than waiting until adoption forces the issue
5. Make AI training for key personnel mandatory rather than optional

The Stanford AI Index 2026 confirms the paradox vividly: Switzerland leads the world with 110.5 AI specialists per 100,000 inhabitants, ranking first globally. Yet only 8% of small companies use AI, compared to 34% of large enterprises. The talent exists within the country, but it is not concentrated in the SMEs that need it most. This is precisely where structured AI outsourcing becomes relevant: access to the existing talent pool without having to win the global race for internal hires.