Contents

• How has AI impacted office-based jobs in the short term?

• Literature review

• Long-term impacts of AI

• What next for occupiers?

• AI's impact on office design

• Conclusion

How has AI impacted office-based jobs in the short term?

As AI continues to develop, its impact on traditional office jobs and occupiers’ future demand for office space will be a key consideration for landlords. In Savills second European office AI spotlight, we utilise empirical evidence of AI’s impact on office-based jobs to outline the potential implications on office demand and building design across Europe.

Internet 2.0

The dotcom boom is the most recent example of a technology boost to the supply side of the economy, during which time, corporate funding to large tech companies rose, internet adoption rates increased, followed by a short-term reduction in hiring. Over the longer term, internet access supported job growth as office workers reskilled to meet the new demands of their augmented roles, with London's office-based employment doubling over the course of 15 years, with a similar trend across other major European cities. We are observing a similar trend in 2025, with the number of new job openings slowing as AI adoption gathers pace.

More recent research from Yale University indicates that ChatGPT has only changed workers’ roles at a slightly faster pace than the arrival of computers or the internet (Chart 2), and the short-term displacement has been greater in the tech industry.

Recent corporate funding

Big tech companies are investing heavily in building and training AI models as they aim to win the arms race as the leader to reshape industries. During 2025 and 2026 alone, Google, Amazon, Microsoft and Meta will spend $750 billion on servers, chips and data centres to power their AI models.

As a result, corporate funding is flowing through to European-headquartered AI and machine learning (ML) companies with €16 billion of venture capital (VC) invested during Q1–Q3 2025, marking a record year, with the largest proportions targeting the defence, drug discovery and financial services industries.

There has also been an increase in the proportion of M&A activity to AI companies as established big tech multinationals seek to scale from faster-growth companies to take advantage of their agility and talent pool, which will increase demand for larger, consolidated office space across European cities.

Recent hiring trends

Employers are considering their recruitment strategies to position their company and workforce most effectively, although any evidence of AI displacing office-based jobs has been limited. In fact, data from Eurostat and Oxford Economics indicates a positive relationship between the AI adoption rate and job growth by office business sector since 2020, suggesting AI adoption is supporting the growth of new jobs, rather than replacing employees.

What is more evident is the impact on hiring younger, more inexperienced workers. In the UK, entry-level job vacancies as a proportion of total job vacancies fell from 29% in 2022 to 22% in 2025, according to data from Adzuna. However, weaker economic growth, geopolitical uncertainty and increases in employers’ national insurance contributions are more significant factors contributing to the slowdown. Employers will seek more experienced employees to utilise AI.

60% of US workers believe that AI will change how they do their current job in the next five years, whereas 36% believe that AI will replace their job in the next five years, according to Fortune. Evidently, workers recognise the need to retrain and invest in their skillsets to meet the new demands of their role.

Literature review

What does empirical research indicate will happen to office-based jobs over the longer term?

Utilising empirical research from various studies, translating the impact of AI into implications for the economy over the longer term could either provide a productivity boost or job displacement:

Productivity boost Office-based workers will be able to utilise AI to produce more with less, focussing more of their time on higher value tasks. Oxford Economics estimates that generative AI could raise labour productivity by 10% over the next decade.

Job displacement AI will automate routine, repetitive tasks, which will reduce demand for labour.

Oxford Economics’ analysis indicates that jobs that include tasks with more exposure to automation are more likely to experience productivity gains, and therefore, service sector economies stand to gain disproportionately from AI adoption over the longer term. Overall, Oxford Economics forecasts that the EU’s office-based employment is expected to grow by 0.4% per annum over the next ten years, while the information and communication sector and the professional, science and tech sector are the sectors with the strongest growth.

The information and communication sector is expected to benefit from a larger productivity boost from AI and is also expected to see the strongest employment growth over the next ten years, as employees’ roles are less automatable, according to Oxford Economics’ analysis.

On the other hand, the finance and insurance sector is expected to be more impacted by job displacement, given employees’ exposure to computing and mathematics-related tasks, for example. Sectors least likely to benefit from AI sit on the production side, including agriculture and construction, which involve a higher proportion of manual work.

Long-term impacts of AI

AI clearly provides an additional level of uncertainty on the future for office-based employment, with many known-unknowns associated with the speed and extent of AI’s impact – here we present the bull vs bear case for each factor.

Efficiency gains

Bull

In a scenario where every company will be adopting AI throughout all of their business lines, AI will be central to collating information to support decision-making, as users will be able to automate processes, reducing the time taken to run basic tasks.

In banking, for example, AI is automating routine tasks like client enquiries and summarising documents so employees can spend more time evaluating loans and building new client relationships. Banks are shifting from AI experimentation to AI implementation/ scale up, in order to differentiate themselves against competitors.

Big tech companies, on the other hand, are developing AI agents that will have autonomy to act on the employee’s behalf in some instances. Whilst some repetitive tasks will be automated, employees will be able to allocate more of their time to focus on complex problem-solving and creative tasks.

Bear

AI’s outputs are built from the quality of its inputs, and in many scenarios, it cannot differentiate what is factually true or false. Human approval is required, and more employee time will be spent checking and correcting AI-generated outputs.

Research from BCG shows that only 26% of companies using AI report that they are generating ‘tangible value’. According to Gartner’s Hype Cycle of AI, GenAI has entered the Trough of Disillusionment, as organisations gain understanding of its potential and limitations. Less than 30% of CEOs are satisfied with their investment return on AI, driven by governance challenges, the need for increasing regulation, and a shortage of skilled professionals. Companies are still exploring what can be achieved, and there is long-term upskilling required across workforces to produce meaningful outputs.

What’s more, CEOs remain cautious on the security of company data within AI systems. Companies are opting to form frameworks around how AI can be used before opting to roll out more extensively.

A further barrier to AI adoption is falling public levels of trust – KPMG’s analysis shows that the proportion of the public who trust AI systems fell to 56%, compared to 63% two years ago. But will society ever fully trust AI-generated outputs with no human accountability? We expect any transition to be a more gradual, long-term shift.

Job displacement

Bull

AI could further reduce the demand for entry-level roles and therefore save companies costs. Already, we have seen the proportion of graduate job vacancies rise. The average graduate salary is currently circa £30,000 in London – equivalent to 150 ChatGPT Plus premium subscriptions. This is even before national insurance contributions, desk costs, and other associated expenses.

Companies could benefit from fewer human errors, faster decision-making, automation of repetitive tasks, and 24/7 availability will support business adoption of AI. Most workers would be required to apply prompt engineering to Large Language Models (LLMs) to ensure they generate accurate, relevant, and desired outputs to steer the AI tool to make effective content.

Bear

Generational divides with adoption of AI will slow the transition. Younger workers are more agile and are often more readily utilising AI to upskill more senior colleagues. Employers will seek agile staff who can easily adopt new technologies.

Goldman Sachs research indicates that AI will improve overall labour productivity and only result in a 6–7% displacement of US jobs. Job displacement is only expected to be short-term for high-skilled jobs as they retrain and find new, augmented roles. For example, analysis from Evident shows that hiring of AI-skilled workers in the banking sector has increased by 13% in the past six months. AI engineers, data scientists, and AI ethics officers are among the types of roles that are likely to be created.

Soft skills remain as important as ever, as employers believe the future of work will remain about collaboration and building professional relationships. According to the World Economic Forum’s Future of Jobs 2025 report, analytical thinking remains the top core skill for employers, with 70% of companies considering it as essential. This is followed by resilience, flexibility and agility, along with leadership and social influence. Creative thinking and motivation and self-awareness rank fourth and fifth, respectively.

Over the next five years, employers anticipate the fastest-growing demand for AI and big data, networks and cybersecurity, technological literacy, creative thinking and resilience skills. Big tech companies are now offering AI and ML certifications for entry-level workers to differentiate themselves. Employers will invest more in upskilling current employees to ensure they are AI-enabled.

In many cases, AI-related job cuts have resulted in falling levels of customer service. Challenger banks have rehired customer service agents, following customer dissatisfaction with AI chatbots. More complex, empathetic customer interactions underlines the importance of human support.

The cost of AI subscriptions will increase over time – big tech companies have invested so heavily in the future growth that they will need to account for their costs in the future, once adoption rates have risen. Some companies will view AI as a ‘nice to have’, rather than committing to paying for premium licences.

AI-related cyberattacks are an additional threat to the growth of AI. More sophisticated phishing/ deepfakes will require further investment in cybersecurity, which is creating new jobs in the sector. Darktrace, for example, expanded into 950 sq m of office space in Munich during 2025.

It is also worth considering that much of the reason for the slowdown in tech hiring during 2024/25 is the over-recruitment during 2022/23. Tech companies are now more strategic around the types of software development roles they are hiring for, which should see more sustainable job growth.

Regulation

Overall, increasing regulations are likely to slow the adoption rate of AI and protect employees and businesses.

The EU AI Act, which came into force in 2024, represents the world’s first comprehensive regulatory framework for AI and has been established in order to protect jobs, promote innovation responsibly, and keep users safe. Regulation is evolving faster than many businesses are able to keep up with. The EU has separately developed a code of practice for AI, as it holds firm on its digital rules, with Google, OpenAI and Mistral having signed the code.

Breaching the EU’s AI Act can carry a fine of as much as 7% of a company's annual sales, or 3% for companies developing AI models. OpenAI recently opened its Brussels office to be closer to regulatory bodies, with a view to focusing on “promoting transparency, safety, and accountability in AI”.

In addition, the RICS has published the first global professional standard for the responsible use of AI in surveying, set to take effect in 2026. The provisions of the new standard include governance and risk management, professional judgement, transparency and client communication and ethical development of AI. The initiative aims to uphold technical and ethical standards across the built environment, ensuring innovation is in the public interest.

A further challenge for Europe’s long-term growth is how its AI regulations compare on a global basis. This could act as a headwind for the growth of AI companies that may opt to expand in countries with less red tape.

What next for occupiers?

Which European occupational markets are positioned most resiliently to AI's growth?

The Savills European Occupier AI Resilience index ranks 22 cities based on their overall occupier resilience to AI, built on a composite scoring system across three weighted pillars: Talent (45%), Business Environment (50%), and Real Estate (5%).

Methodology

• The Talent pillar captures the depth, cost, and mobility of AI-skilled professionals. Variables include average labour cost of a software development engineer (10% weighting, source: Levels.fyi), total number of software engineers (25% weighting, source: CEPS’ “Solving Europe’s AI Talent Equation”) and AI talent mobility (weighting 10%, source: OECD’s 2024 data on net inflow per 10,000 LinkedIn members with AI skills – a proxy for how attractive each city is to mobile, high-skilled professionals).

• The Business Environment pillar reflects the financial and regulatory conditions that shape the viability of AI enterprises. Variables include VC investment in AI and ML over the last ten years (30% weighting, source: Pitchbook), tax burden, which aggregates corporate tax rates and employer social contributions (10% weighting, source: Tax Foundation, OECD) and ease of doing business which captures regulatory quality, infrastructure, and business friendliness (10% weighting, source: World Bank’s B-READY Index).

• The Real Estate pillar considers the prime office rents for European cities (5% weighting, source: Savills).

Together, these seven indicators form a forward-looking benchmark of AI occupier demand resilience across Europe at the city level.

Savills Top European AI cities

1) London

London’s dominance stems from its sheer scale and maturity as a tech ecosystem. The city benefits from deep pools of VC – raising €26.5 billion in AI/ML VC over the past decade, and a highly skilled workforce of approximately 120,000 software engineers. Its global reputation, world-class universities, and regulatory environment continue to attract top-tier talent and investors.

Established AI companies are expanding across London, with Synthesia signing for 2,000 sq m at Triton Square, upsizing by four times its previous office footprint. ElevenLabs, Salesforce and PhysicsX have also signed for office space in London over the last twelve months.

2) Berlin

Berlin’s success is rooted in its affordability, talent density, and strong public support for innovation. It hosts 25 unicorn founder factories, which have spun out 165 startups, many of which remain in the city. With around 55,000 engineers and a high net AI talent inflow, Berlin combines scale with cost-efficiency. Government-backed programs, including the Berlin Start-up Scholarship and Investitionsbank Berlin, have injected over €250 million into early-stage ventures.

3) Munich

Munich has raised over €3 billion of AI VC funding over the last ten years, with startups flourishing as a result of being located in close proximity to established tech clusters. With over 40,000 software engineers located in the city, Darktrace, Snowflake and OpenAI have established offices.

4) Paris

Paris has emerged as a major European AI hub, home to over 1,000 AI startups. It benefits from elite engineering schools (like ENS and Centrale), strong R&D investment, and generous government support. President Macron has also announced €109 billion of investment in AI in the coming years at the Paris AI Action Summit 2025, despite domestic political uncertainty. With around 70,000 engineers and €9.9 billion in VC raised since 2015, Paris combines scale, policy support, and innovation density.

Larger US tech companies are now seeking to increase exposure to the French market, with OpenAI opening its first mainland European office in Paris in 2024 to be closer to European clients. Mistral AI, headquartered in Paris, reached an €11.7 billion valuation in 2025 and has announced plans to invest in its first French data centre, reducing its reliance on US cloud computing service providers.

5) Dublin

Dublin’s rise is driven by its strong positive talent migration and business-friendly regulatory environment. It’s a key European base for global tech giants like Google, Meta, and Microsoft, which has helped build a robust local talent pool. Ireland’s corporate tax regime and streamlined business setup processes make Dublin particularly attractive for startups and scale-ups looking to expand into Europe. While its engineer count is lower, its affordability and international connectivity give it a competitive edge.

6) Stockholm

Stockholm’s strength lies in its deep-rooted innovation culture and globally recognised startup ecosystem. Often dubbed the “Unicorn Factory”, it has produced tech giants like Spotify, Klarna and Skype. The city benefits from a world-class education system, with institutions like KTH and Uppsala University feeding the talent pipeline, and a digitally savvy population supported by one of the largest fibre networks in the world. Stockholm also stands out for its ethical approach to AI and its collaborative tech culture, which encourages knowledge sharing and cross-sector innovation.

AI's impact on office design

How will AI's rising energy requirements changing building design?

Digital connectivity will become even more important for occupiers, and offices will differentiate themselves with the speed and reliability of internet connections. Comms rooms have traditionally accounted for approximately 30–40% of an office building’s operational use, given the power requirements, heat generated and cooling systems required to operate. Although servers have densified significantly in recent years, heat radiation and air conditioning are still significant challenges to reducing energy usage.

Planning, power and procurement are challenges to the speed of development for data centres. In the UK, data centres have recently been classified as critical national infrastructure, which will receive additional government support in the case of cyber attacks, IT outages, or extreme weather, in order to minimise disruption. The chief executive of the National Grid has announced that data centre power usage would increase sixfold over the next decade.

In some instances, landlords are demanding that new tenants relocate their comms rooms off-site into data centres, reducing the building's overall energy use and improving the operational energy performance. Given lower operational energy usage, landlords can remove additional cooling systems that are no longer required.

Floor loading requirements for server rooms are significantly higher than those for standard office spaces, which must account for the weight of the racks. Secondary, non-CBD office buildings with compromised floor to ceiling heights, higher floor loading capacity, and good power availability are now being earmarked for conversion to data centres. Building size requirements for data centres have reduced, and more landlords are announcing plans to convert office buildings into data centres to keep up with growing demand.

How does the workplace need to adjust to AI?

Relationship building will become more important, and the strength of these relationships will differentiate workers. Office space will increasingly be focussed around in-person interaction and collaboration, in order for companies to maximise their impact.

The workplace needs to adjust to the new requirements of activity-based working. Overall, the average number of meetings has increased by almost three times over the last five years, given the rise of video conferencing calls, with employers concerned about stagnating productivity levels.

The office is becoming increasingly focussed around smaller team interactions. We are seeing more demand for breakout areas designed solely for in-person interactions for core teams, designed for two to six people. More focus is being placed on creating ‘chance’ interactions in a more creative setting for workers to generate new ideas. For wider team meetings, there is increased demand for larger meeting rooms with dial-in facilities.

AI is likely to increase hybrid working trends over the long term, although it is too early to have seen any evidence of this so far. Businesses are still adjusting to remote working practices and reflecting this in their office footprint. We had already observed an ongoing shift to more private meeting rooms/ focus pods for private phone calls and focussed work. With the growth of voice recognition in the AI industry, this is likely to increase in the demand for more private booths in offices.

Indeed, the British Council for Offices’ latest report indicates that the office utilisation rate benchmark of 80% is no longer fit for purpose, and now 66% is a more realistic figure.

Given the new demand for a wider variety of workspaces, companies will find their existing space is no longer sufficient.

Conclusion

AI is an additional long-term risk to the demand for office-based jobs and office demand. However, we believe any changes to office-based jobs are likely to be incremental over the long term as jobs are augmented rather than automated.

Until the AI adoption cycle has fully played out, the potential labour market disruption (including which jobs are likely to be displaced) will remain an open question. In the short to medium term, AI adoption rates are likely to increase, with more employee time spent checking AI-generated output. The level of regulation and public trust will also contribute to determining the speed and extent of AI's impact, particularly in Europe.

Multinationals will seek to expand in European cities with AI-enabled workforces and established AI clusters, including London, Berlin, Munich, Paris, Dublin and Stockholm. Employers will invest more in upskilling current employees to ensure they are able to utilise AI tools.

Office workers will ultimately need retraining to adapt to new job requirements over the longer term. Job displacement risks are higher for entry-level jobs and more repetitive roles. Skills required across all business sectors will be more focussed on utilising existing AI tools, freeing up more time to be allocated to interactive, people-focussed and creative roles.

Office buildings will need to adapt. For the workplace, more emphasis will be on the benefits of collaboration and strengthening relationships through smaller team interactions. Employees will need a wider choice of workspaces to accommodate more varied working styles, including open breakout areas, larger meeting rooms, meeting booths and focus pods, which will increase average workspace required per employee.

Building design will also adjust to reflect the increased energy requirements from AI. Landlords are seeking to remove server rooms and are requiring tenants to move to cloud-based storage to do so. Repurposing opportunities for investors will come to the fore, as secondary offices with power availability are converted to data centres to support the growing demand for data storage.