A Swiss AI initiative is showing how state-backed computing and long-term investment can challenge corporate AI while protecting creators, lowering costs, and keeping technology focused on public needs first.
The argument is not that AI itself creates every harm linked to technology, but how AI is financed, trained, and deployed when private profit becomes the supreme goal.
Public investment offers another path, models built with clearer rules, licensed or filtered data, renewable energy and access that serves researchers, governments, businesses and citizens.
Switzerland’s Apertus Gives Public AI a Chance
The Swiss AI initiative has turned public AI from an idea into infrastructure that people can use. At the center of that effort is a Swiss AI model built to be open, inspectable and available for reuse.
The Apertus language model was trained at a scale that is large enough to be useful without trying to chase every commercial benchmark.
That matters because corporate AI development often rewards size, speed, and market value. Public AI can follow different logic, focused on usefulness, cost, legal safeguards, and long-term public value instead of pushing users toward products designed mainly to increase private returns.
The Swiss artificial intelligence (AI) approach also shows why public control can matter when training data comes from creators, publishers and other rights holders.
Apertus took extra steps to respect data opt-outs under the EU AI Act. Its developers went back to original sources used in common training datasets and removed more than 2.5 million documents. That does not solve every copyright concern, but it shows how an AI project can treat creators’ choices as part of the design rather than a barrier to growth.
The Swiss AI platform can compete on trust as much as technical performance, but cost is another part of the equation. Apertus used about 6 million Graphics Processing Unit (GPU) hours in its final training run, with total experimentation estimated at around 10 million GPU hours. The commercial value of that computing is between $36 million and $100 million.
The Swiss AI initiative also benefited from millions of GPU hours made available through Switzerland’s national computing infrastructure.
The stated goals include building trustworthy AI, protecting Switzerland’s digital sovereignty and creating technology that could serve industry and public administration, putting public control, national capability, and access at the center of the project.
The Swiss AI Alps supercomputer gives this policy something corporate alternatives cannot easily copy publicly backed computing capacity tied to research goals. The Alps supercomputer was built through cooperation between science, government, and industry, and is designed for very large research workloads.
“Alps is an expression of our vision of a future characterized by knowledge and progress,” said Federal Councilor, Guy Parmelin, at its inauguration.
The Swiss AI model also has an environmental argument behind it.
Just Apertus was trained using renewable Swiss hydropower, it does not make AI free of environmental costs. Au contrary, it shows that governments can make energy choices part of their AI strategy where a public system does not have to maximize computing use simply because a larger model may score better on a benchmark.
Switzerland’s Money, Talent, and Reason to Just Push
Swiss deep tech start-ups raised $2.6 billion last year, according to the Swiss Deep Tech Report 2026. Around 63% of the country’s venture capital (VC) goes into deep tech, a larger share than in any other country cited by the report. AI and machine learning now account for 25% of Swiss start-ups covered in the sector.
The case for stronger public AI investment becomes more credible when it is placed inside deep tech nation Switzerland.
The strength of deep tech Switzerland gives Apertus access to something that many public technology projects struggle to build: a wider ecosystem of researchers, engineers, universities, investors and new companies.
Since 2023, ETH Zurich and EPFL Lausanne universities have produced 24 and 16 deep tech spinouts respectively, as per the report. Their role also matters for public AI because Apertus itself grew from the same research environment.
The Swiss AI initiative can use that ecosystem to keep talent, computing knowledge and AI development closer to public institutions. But the investment picture is also important.
Overseas, investors provide 88% of Swiss deep tech funding in rounds above $100 million, highlighting that international capital already sees value in the country’s research base.
“For the first time, the companies spinning out of ETH and EPFL are staying, scaling and attracting serious capital,” said Jean-Philippe Fricker of Cerebras Systems.
The Swiss AI platform could benefit from the same confidence if public investment remains steady, and access maintains its current affordability.
Public AI does not need to beat every private model on every test to succeed. It needs to be good enough for real work while offering benefits that closed systems cannot easily provide visible training choices, open code, public accountability, lower barriers to access, and clearer rules for reuse.
The Swiss artificial intelligence (AI) model can also support public administration without forcing governments to send sensitive work through outside commercial systems.
AI can become basic digital infrastructure, closer to a library, a university network, or a public utility. The goal is not to eliminate private AI, but to ensure that citizens and institutions have a credible option when private systems do not meet public needs.
The Swiss AI model gives Switzerland a foundation it can improve instead of renting intelligence from a small group of global technology companies.
The Swiss AI initiative is important because the first version of a public model is not the end of the race, it is the beginning of a long investment cycle.
If Switzerland keeps funding computing capacity, research, energy efficiency and public access, Apertus can become stronger without giving up the principles that make it different. That could help prove that openness and public value are not weaknesses in AI development.
A public system does not have to copy the corporate AI race to matter. If it stays transparent, affordable, sustainable, and useful, it can protect creators, serve society and still compete for the people and institutions that want AI built around public needs.
The Swiss AI initiative could then offer a wider lesson, public AI can win when governments are willing to invest in it as infrastructure, not treat it as a side project.
Inside Telecom provides you with an extensive list of content covering all aspects of the tech industry. Keep an eye on our Intelligent Tech sections to stay informed and up-to-date with our daily articles.
