The ‘fabs’ themselves are strictly custom-built, usually by their occupiers. As such, we have split this analysis into two parts. Firstly, we look at typical densities for employment in the manufacturing sector. We consider space taken by the fabs themselves and the direct employment generated by the industry, which s attributed to logistics demand.

The second component of our demand estimation examines total job creation, both direct and indirect and attributes household demand to logistics. While many workers would have found employment elsewhere, demand induced by semiconductor employment is likely to occur in specific regions creating excess demand in certain areas like that experience in areas where TSMC or Intel have previously set up. In Arizona, for example, there has been a significant increase in household construction due to the boost in employment TSMC has brought to the area, and logistics demand will consequentially rise in response to population growth in a given area. Furthermore, competition for labour created by a growing semiconductor sector will drive significantly above-average wages. Employees in the industry will have higher disposable incomes and, consequentially, a higher propensity to consume.

The SIA suggests that a one-off $50bn federal incentive program could support the construction of an additional ten fabs in the US that would not have been built otherwise. The SIA suggests this would increase direct employment by 42,000 jobs from 277,000 to 319,000 in the US. Assuming a jobs multiplier from earlier research by the industry body, which found a 6.7 jobs multiplier, the semiconductor industry would support an additional 280,000 jobs in this scenario. Of these, 101,500 (36%) would be indirect employment in industry and services adjacent to the semiconductor industry and 137,300 jobs from induced employment.

WE MAKE A NUMBER OF ASSUMPTIONS IN THIS ANALYSIS:

• The EU successfully increases its share of semiconductor production as set out in the EU Chips Act.

• For household demand, we assume an average household size of 2.3 persons, in line with the EU average.

• We assume 6.4 sq m of demand per household for logistics space generated by household consumption.

• We assume the 6.7 jobs multiplier for the semiconductor industry set out by the SIA.

FORECASTED DEMAND

If we assume that the EU Chips Act has a similar outcome to the $50bn investment proposed by the SIA, and ten fabs are built across Europe, we expect to see an additional 38,000 roles in the sector. This equates to 232,000 sq m of demand. Notably, this represents the lower bound in terms of the size of these units, with many fabs typically being significantly larger. This volume of demand in isolation is not particularly significant in the context of the overall European market, which has seen an average of 29.2 million sq m of take-up each year since 2014.

Savills believes the more substantial impact would come through additional indirect employment generated, which could drive up to 4.9 million sq m of take-up in response. We assume a conservative share of these roles are in complementary manufacturing and logistics. While this is in and of itself not a massive increase in take-up, in tandem with other tailwinds in the logistics market, it could contribute to downward pressure on the vacancy rate.

Our upper bound calculation in this area will assume that the EU successfully doubles the size of the semiconductor industry. Production will have to more than double to achieve the 20% of global output outlined by the EU Chips Act, as other parts of the world also look to expand their output. Taking the 200,000 workers in the sector quoted by the ESI and assuming that this number will double over the period to reflect the substantial increase required, which translates to an additional 244,444 workers to reach a 20% share.

Extrapolating from the SIA’s figures, this suggests that 58 fabs would need to be constructed by 2030. For context, in 2023, a record 33 fabs are under construction globally as semiconductor supply begins to ramp up. Seventeen new fabs are expected to complete between 2021 and 2023, a record for the region. In this context, while these ambitions, like the EU Chips Act itself, are lofty, it is not impossible for this quantum of space to be delivered.

This would require up to 1.35 million sq m of space for fabs alone. Crucially semiconductors would directly compete with other manufacturers and logistics operators for the development land required. In terms of the indirect employment driven by the sector, we would expect to see between 2.4 million sq m and 4.9 million sq m of logistics demand arising from this level of development.

HOUSEHOLD CONSUMPTION

In addition to the demand driven by the fabs and the ancillary businesses that support them, the employment that the industry will generate will lead to further demand for consumer goods from the households formed by employees.

Analysis by Savills suggests that each additional household will create a requirement for 6.4 sq m of logistics space. In the context of the 280,000 additional jobs in line with the ESI’s forecasts, this would generate some 847,000 sq m of demand in the EU. Expanding on this with our successful EU Chips Act scenario, we would expect to see up to 4.5 million sq m of logistics demand from the 1.6 million additional jobs created through direct, indirect and induced employment as a result of the strong growth required to reach this level of output. Notably, jobs in the semiconductor industry will command wages above local economies’ average wages, giving these households higher disposable incomes and levels of consumption.

In combination with the direct demand from semiconductor firms and the indirect demand stemming from their output, this would boost demand by up to 10.8 million sq m. This would occur over the course of the next seven years and would provide a boost to overall logistics demand and leasing activity, driving down the vacancy rate and supporting rental growth in the long term.

We assume that take-up reverts to its pre-pandemic long-run trend between 2024 to 2030. Taking the five-year average of 2014 to 2019 of 25.62 million sq m of take-up and apply the compound annual growth rate for the period (7.1%), we can forecast take-up, assuming that the eCommerce boom of 2020–2022 was an anomaly. This is likely to understate take-up as e-commerce’s rise represents a structural change in the market, so we expect take-up to be higher than this. Distributing semiconductor-induced take-up evenly across the period, we would expect to see an additional 1.55 million sq m of demand per annum. This accounts for an average of 4.4% of overall European logistics demand.

_{Read the articles within Spotlight: Semiconductors and the logistics sector below.}