1. France has emerged as the leading growth engine for European data centers.

The French government has stated that it will seek to cut bureaucracy, including on planning permission, AI regulation, access to energy etc.
President Macron noted that France produces more electricity than it consumes, making the country an attractive destination for energy-intensive AI companies.
France sources ca 65-70% of its energy from nuclear sources (with renewables representing ca 20%) - the most of any European country. Nuclear energy has many advantages: very concentrated levels of baseload power, low carbon footprint, and is typically bundled with land/water/ security.
Also, Telco connectivity (incl submarine cables) is excellent: France - and its DCs - are directly connected to North America, the UK, Africa and Asia via a network of submarine cables.

Below: France offers relatively cheap electricity compared to other European countries

2. French DC market to grow from 0.8 GW (end of 2025a) to 6.7 GW (by Dec 2035).

Below: companies to gain from data center growth. Obviously, all French companies will benefit, such as Schneider, Legrand, Engie, Eiffage etcetera

3. Legrand confirms that current growth in data centers remains very strong and assured it will not end in 2026.

Data center total addressable market for Legrand in the medium term is around €400 billion, says JPMorgan.

4. AI makes the IT labor-intensive consulting models obsolete.

Accenture, the global consulting and IT services giant, got thrashed after its earnings report made investors ask the ugliest AI question: what if the technology that consultants sell also eats the consulting model?
The stock fell 18% to $127.14. Fiscal Q3 adjusted EPS was $3.80, above estimates, while revenue came in at $18.7B, below the roughly $18.8B forecast. New bookings fell 3% to $19.3B, and the company narrowed full-year revenue growth guidance to 3%-4%.
The quarter was not a disaster on the surface, which is what makes the selloff interesting. The market is saying the old labor-heavy consulting machine has a credibility problem if clients slow discretionary spending while AI agents make more work automatable. Accenture is still buying its way deeper into AI and cybersecurity, but investors want proof that those deals replace billable-hours pressure instead.
The stock is deeply oversold but potentially a very sharp falling knife.

5. Inference versus training.

As AI workloads are shifting from training toward inference, analists expect inference to outgrow training through 2029.

Not a subscriber yet?