Demand for datacenter capacity in Europe sees busiest Q2 ever

The international economic system could be affected by inflation and low progress, however that hasn’t stopped the primary half of 2023 from being the busiest on report for datacenter takeup in Europe, based on new analysis.

A report [PDF] from business actual property outfit JLL claims there was report demand for datacenter capacity through the interval from April to June this 12 months alone, with 114 MW of take-up throughout Europe’s main markets together with Frankfurt, London, Amsterdam, Paris and Dublin – referred to by the charming acronym of FLAP-D.

London stays the most important market with a complete of 902 MW, based on JLL, amounting to a 35 % share of the overall Tier 1 capacity. However, right here there was a fall-off in the amount of recent capacity added up to now this 12 months, at solely 7 MW.

In distinction, Frankfurt noticed 44 MW of extra takeup through the second quarter, bringing town’s complete for the 12 months up to now to 80 MW, in comparability to only 26 MW on the identical level final 12 months.

Capacity in Amsterdam sits at 458 MW, with 10 MW added in Q2, however there was no important new provide added to this market since a moratorium on new datacenter growth ended in 2020.

Paris has seen 24 MW of recent capacity added up to now in 2023, whereas Dublin gained a further 12 MW, growing the general market dimension by 6 % to 199 MW, JLL stated.

According to the report, at the least a few of this wave of recent uptake could be attributed to pent-up demand. The final couple of years noticed important ranges of pre-leasing as prospects aimed to safe new datacenter area whereas it was nonetheless being developed, JLL stated, whereas the availability chain points that had been seen throughout the trade led to completion dates being pushed again.

The report additionally notes that the emergence of generative AI has been having an influence on datacenters, an impact additionally beforehand commented on by analysis outfit Omdia. In reality, the excessive compute energy necessities wanted for the coaching of AI fashions is driving a elementary change in the very design of datacenters, JLL claimed.

It states that high-performance GPUs have been added to the compute infrastructure in order to fulfill the necessities of coaching AI fashions, and these are pushing the necessity for liquid cooling fairly than air cooling.

JLL’s Head of EMEA Datacenter Transactions Tom Glover observes that infrastructure for coaching AI fashions is much less reliant on latency, so proximity to the top person could also be much less of a consideration.

This could result in some datacenters being strategically positioned the place entry to renewable vitality, out there land and water for cooling just isn’t a difficulty, he claimed. However, this would appear to suggest such a datacenter being constructed from the outset to be largely devoted to operating AI coaching workloads.

• Our AI behavior is already altering the way in which we construct datacenters
• Why securing East-West community visitors is so vital – and the way it may be finished
• Virginia industrial park needs to energy DCs with mini nuclear reactors, clear hydrogen
• Tesla is trying for folks to construct ‘1st of its variety Data Centers’

Glover stated that non-public fairness companies are prepared to speculate based mostly on the expansion potential of AI. He cited Blackstone as earmarking properties constructed to deal with the rising information wants from a surge in AI, whereas DigitalBridge has introduced a fund of $1.2 billion geared in direction of AI funding.

However, a separate report earlier this 12 months claimed that the European datacenter trade was going through challenges in assembly the rising demand for capacity, because of points equivalent to supplies and heavy gear to construct websites being in quick provide.

It said that there was an inadequate pipeline of datacenter growth deliberate in Europe over the subsequent three years to fulfill the forecast enhance in demand. ®