What Swiss SMEs Can Learn From China's Involution Culture

What involution (内卷) actually means

The term 内卷 - involution - describes a form of competition in China that consumes itself: companies invest ever more effort without the industry as a whole becoming more profitable. Rhodium Group documents this mechanism in numbers: since January 2022, China's producer price index has fallen by 5.4 percent - margins keep shrinking despite rising productivity. Involution is therefore not a buzzword but a structural problem that forces entire industries into a race with no winners.

Involution is most visible in China's electric vehicle industry. Between 2017 and 2024, the average profit margin of EV manufacturers fell from 7.8 to 4.3 percent. The reason: to stay relevant, manufacturers had to update their models on weekly rather than annual cycles - any delay meant lost market share. That speed came at a price: between 2018 and 2025, 400 of 500 Chinese EV manufacturers disappeared from the market.

The downside: when competition turns into self-destruction

Beijing itself now recognises involution as a problem. Within two quarters of 2025, food-delivery platforms burned through the equivalent of more than 100 billion yuan in a pure market-share price war - without any company achieving lasting profitability. Bruegel describes the result as 'growth without profits': markups are falling across industries while the number of zombie firms, kept alive only by cheap credit, keeps rising. Michael Pettis of the Carnegie Endowment frames this as concentrated overcapacity - Beijing even had to set up a fund of 50 billion renminbi in the polysilicon sector to force market consolidation.

Switzerland's mirror problem: too perfect, too slow

Swiss SMEs face the opposite problem. Instead of drowning in ruinous speed, many stay stuck in pilot mode: a large share of Swiss companies experiment with AI but rarely make the shift from pilot project to production use. Swiss quality standards are a genuine competitive advantage - but they become a brake when confused with endless approval loops and multi-year planning horizons.

Three lessons for Swiss SMEs

• Weekly rather than annual iteration: small, frequent improvement steps instead of one big, perfect launch
• Open-source speed: build on open models that evolve fast, rather than waiting for a closed proprietary development
• The right measure: adopt the iteration culture without sliding into price dumping or overcapacity

Lesson 1: Weekly iteration beats perfect planning

Chinese EV manufacturers learned to make product updates, pricing decisions and supply-chain choices on weekly cycles instead of annual plans. For Swiss SMEs this doesn't mean acting rashly - it means radically shortening the learning loop. Model agility matters more than the one-time, perfect choice of an AI model or vendor. Companies that test, measure and adjust weekly learn faster than even the most careful annual plan could ever allow.

Lesson 2: Open-source speed instead of proprietary moats

Models such as DeepSeek or Moonshot's Kimi show how open development shortens innovation cycles: improvements spread immediately instead of maturing for years inside closed systems. This pace can be harnessed outside China too - not by copying individual models, but by adopting the same openness towards fast exchange and iteration.

Lesson 3: Finding the right measure

Not all speed is healthy - involution itself is the opposite of a sustainable model. The real skill lies in raising iteration frequency without sacrificing margin or quality. AI strategy for Swiss SMEs outlines three common mistakes in the global race - including the CEO paralysis that arises when companies analyse for too long instead of testing before they invest at all.

AI outsourcing as a way to gain speed without overhead

The reason many Swiss SMEs fail to iterate faster is rarely a lack of will but a lack of capacity: building an in-house AI team costs time and money that mid-sized companies rarely have to spare. A recent study on European SMEs shows the potential when AI is used consistently rather than sporadically: companies with integrated AI use achieved productivity gains of up to 47 percent. AI outsourcing solves exactly this dilemma - it delivers the iteration speed of external specialist teams without requiring an SME to build its own structures.

The way forward

1. Set an iteration rhythm: weekly or bi-weekly instead of annual
2. Choose one clearly scoped process for the first test run
3. Use external AI capacity to build speed without scaling in-house teams
4. Maintain price and margin discipline - speed is a means, not an end in itself

Involution is both a warning and a lesson. Swiss SMEs don't have to choose between Swiss quality and Chinese speed - they can adapt the iteration discipline without abandoning the principles that made their brands strong in the first place.