The Decisive | Season 6 | Ep.7 - How AI Investment Is Influencing Labor Dynamics

View Full Transcript

Kristen Hallam:  

Today, we're focusing on artificial intelligence, a sector experiencing remarkable growth fueled by investments from both private and public sources. With government support in select economies, the momentum behind AI innovation shows no signs of slowing down. As AI continues to reshape industries, it brings with it significant challenges, especially regarding labor dynamics. While much of the discussion has focused on potentially negative impacts on the labor force, we see opportunities and areas for growth. Over the next hour, we'll examine how these trends are expected to evolve through 2026 and into 2027. We'll explore how AI investment is transforming occupational employment patterns, the outcome priorities companies are focusing on as they adopt AI, the level of autonomous AI deployment and its impact on labor demand and the key factors constraining broader implementation of AI. Let me introduce you to our experts now. Sophie Malin, Principal Economist, Pricing and Purchasing; Pollyanna De Lima, Economics Associate Director and our Purchasing Managers Index team; and Alisa Strobel, Principal Economist, Sub-Saharan Africa. So, I'd like to start with the big picture. How prevalent is AI in businesses today? And where do we see the highest rates of adoption, both by sector and geography, Polly?

Pollyanna de Lima:  

Thanks, Kristen. Before I answer that, let me provide some context. So, we have our PMI data extracted from monthly business surveys that track current economic conditions by asking companies about key measures like output, employment and inflation, and they cover over 40 economies worldwide. Now beyond our standard insights, our extensive survey capability allows us to focus on specific themes of economic importance. For instance, we've conducted a special survey on artificial intelligence to assess firms' adoption plans and expectations worldwide, providing unique insights into how emerging technologies are reshaping business landscapes across different sectors and regions. What we have seen in the latest results is that the proportion of firms investing in AI rose in all countries and sectors that we surveyed, and they are generally looking to increase further over 2026. Confidence on this strengthened since 2024 when we last conducted the survey in all countries and sectors. We can also break down the data by company size, which says a consistent story. past and future investments in AI rose at small, medium and large companies. By sector, we see a slightly higher proportion of service providers investing in AI than manufacturers. Within services, financial intermediation is leading in terms of investments over the past 12 months, while electrical goods producers and mechanical engineering topped the rankings for manufacturers. By geography, the AI data showed Germany as the standout leader with 53% of private sector companies investing in AI. The U.K. followed closely behind at 41% and then Spain at 33%.

Kristen Hallam:  

Thanks, Polly. So how much do we expect AI investment by private sector firms to grow this year?

Pollyanna de Lima:

That's an excellent question, and it ties perfectly into what I was mentioning earlier. The results show that these are not just one-off investments and a massive driver behind the sustained growth is likely the wave of new government incentives. We see, for example, 64% of private sector firms in Germany planning further AI investments during 2026. That confidence is potentially supported by state backing like the new EUR 130 billion investment launched by the Federal Ministry of Economic Affairs and KfW as well as generous R&D grants and tax credits under the Research Allowance Act. Similarly, more than half of all survey participants in the U.K. intends to allocate capital towards artificial intelligence. This private capital is being directly encouraged by the government's recent AI opportunities action plan, which is rolling out AI growth zones with streamlined planning and clean power provision. Globally, 27% of manufacturers reported plans to invest in AI over the course of 2026, and that figure stands at 31% for service providers. In selected economies, we're seeing governments fuel this adoption by streamlining data center approvals, offering tax exemptions and providing direct subsidies to ensure that firms have both the infrastructure and the financial backing they need to scale these technologies.

Kristen Hallam:

Okay. So AI is moving fast. Let's bring it back to the labor market. What does this mean for jobs? Sophie, how disruptive has this been or will this be to the labor market?

Sophie Malin:

Well, there's good and bad news. So in our 2024 survey, the survey that Polly was alluding to, we actually saw a net positive impact on employment. So more firms were increasing headcount due to AI investment than we were reducing it. But in our 2025 update, that shifted slightly. So, it's shifted kind of to a small net reduction. So, on net, 5% of firms were reducing employment because of AI. Now that picture does vary by firm size. So larger firms are on net reducing headcount, whereas smaller and medium-sized firms are still expanding employment. So, there is a kind of a clear difference in how AI is affecting jobs depending on the type of firm. Now back to some positive news. So, looking ahead, firms expect -- so this is their expectations for the next 12 months -- there is actually a smaller net share of firms reducing employment. So really kind of when we look at that kind of like net impact, the number of firms reducing headcount specifically because of AI investment has been relatively limited, and we expect it to remain so. So, there's no kind of like downward spiral at least from what we're seeing in our data. And that's really important because when you kind of look at the sensationalist headlines that we've seen, it's all been about kind of widespread AI-driven layoffs. We're not really kind of seeing that play out in our data, at least not yet. So, to give kind of a bit more kind of contextualization, there's other measures of kind of headcount, and more general kind of overall headcount trends. We can look at the S&P Global top 1,200 companies. And on net, 60% of these firms have reduced employment over the past year. So that's 60% compared to 5% when it's related to AI. So that gives a bit more kind of important context that it's more of this downward pressure on jobs is coming from broader economic environment rather than AI investment itself. So -- and it's great because this is exactly what our poll said. But what we are really seeing is that most firms are just maintaining their current workforce levels. Now all of what I've just said is very much kind of a near-term picture. Right now, firms are primarily focused on AI, improving efficiency, improving productivity rather than kind of active headcount reduction. But as we see kind of firms mature in their AI journey, there will likely be more of a focus on how AI can reshape labor requirements.

Kristen Hallam:

Thanks, Sophie. That's really interesting because we're not seeing big headline layoffs, but we're also not seeing a hiring boom. Why aren't employers taking action one way or the other?

Sophie Malin:

Yes. This really kind of comes down to a mix of the current economic backdrop, but also uncertainty around AI and its impacts, right? So, when we look at the current labor market dynamics, labor demand has really steadily weakened over the past few years as it's been activity in the economy has very much been weighed down by higher interest rates, ongoing political and trade uncertainty. So, firms just don't really have a really strong incentive to ramp up hiring at the moment. But at the same time, the impact of AI is quite hard to quantify. So, a lot of firms are using AI in quite ad hoc ways, small-scale kind of projects rather than really enterprise-wide strategies. So, without kind of that clear kind of understanding on what the returns on AI investment really are, firms just don't have enough clarity to make big decisions either way. So either increasing hiring or decreasing hiring. And as I said already, like it really depends where firms are in their AI journey. So many very much in those kind of early adoption phases. That's really where the main focus is kind of productivity gains where we might see a reduction in the need for hiring. But as we see AI more embedded, there's a potential that we might then see stronger output growth because of AI, which could lead potentially to more hiring, but that's a much more kind of later stage effect and it's much less certain about that. So a lot of firms are really kind of waiting and seeing what is going to be the true impact of their AI investments before they make kind of more significant changes to their hiring. And there are constraints on both sides to consider as well. So in areas where labor demand is picking up. So if we think enterprise architects, technical kind of AI implementation roles, there's real skill shortages. So that kind of limits as well how much they can hire. But then on the downside, we saw companies really learn a lesson of how hard it was to rehire after kind of cutting staff during the pandemic. So, there's been of late a much higher kind of like tendency to hoard labor recently rather than large layoffs. So I think that's also playing a little bit into decisions. So really, firms are just adjusting more cautiously, holding back on hiring rather than making any huge moves in either direction.

Kristen Hallam:

All right. Thanks, Sophie. And when adoption does spread into new sectors, it usually shows up first as changes in day-to-day tasks. So, let's talk about skills. Which new skills are emerging or in demand as a result of AI, Sophie, which new jobs will be created linked to AI?

Sophie Malin:

Yes, absolutely. So yes, we are seeing some roles made redundant. So those are particularly those with kind of routine and repetitive tasks that are much easier to automate to think towards your kind of administration, customer service, even kind of parts or software development. But at the same time, AI is reshaping where hiring demand is emerging. So based on our surveys, a lot of job growth is concentrated in roles that are building or enabling AI. So that includes areas like AI, machine learning, research and development, AI product development and AI software engineering. We're also seeing kind of stronger demand in data-related roles, so things like data analytics and engineering because whilst also having the AI firms also need the right kind of data infrastructure to really realize the gains from AI at scale. And then, of course, with any major kind of technological investment, there's also greater demand for operational support. So IT service management, enterprise architects and the teams that are really responsible for running and maintaining these systems. So really kind of the job creation we're seeing is about strengthening the broader AI ecosystems, bringing together technical expertise, the data capabilities and the operational support needed to facilitate the use of AI.

Kristen Hallam:

And that ties into something else we see in the data, which is a real gap between what big companies can do with AI and what small companies can do. What can you tell us about that, Sophie?

Sophie Malin:  

Yes. So we are seeing a clear gap between the share of firms investing in AI across large companies compared to that of small and medium enterprises. And that really kind of comes down to resources. So larger firms have more capacity to invest. So in the technology itself, creating the data infrastructure behind it or being able to hire and train the talent needed to implement and manage AI capabilities. They're also more likely to move beyond kind of those ad hoc projects that I was talking about, implement AI kind of across the business. So using it across various capacities from, let's say, recruitment to sales and marketing and even database management to produce much more widespread productivity gains. But also then large firms are able to build more tailored solutions because of the infrastructure and resource advantage that they kind of have. So it then also enables them to like produce kind of more targeted productivity gains. Smaller firms, however, they often have to rely on more kind of off-the-shelf tools. They've got less ability to invest in the data and systems to fully integrate AI into their operations. So they're using it more for automation. So let's say, minute taking, document production or even kind of social media marketing. So really, large firms are just better positioned to be able to build out more kind of substantial and tailored AI ecosystems, while smaller firms are adopting it more gradually and in a more kind of limited way. So that's really what's leading to that smaller relative share of small and medium businesses investing in AI when compared to larger firms.

Kristen Hallam: 

All right. Thanks, Sophie. Now the AI adoption, labor impact picture looks really different depending on where you are. Developed markets like Germany and the U.K., which you talked about, Polly, are forecasting a net negative impact on employment and emerging markets like India and Brazil are highly optimistic about the employment impact. Why is AI viewed as a massive job creator in emerging markets while being used for labor consolidation in developed ones?

Pollyanna de Lima:

That's an interesting aspect to consider. The difference comes down to a few key factors: labor costs, the stage of AI adoption and the economic context. In developed markets like Germany and the U.K., wages are high and labor is expensive. This makes AI an attractive tool for labor substitution, automating routine tasks and doing more with fewer workers. Our survey supports this. Nearly 50% of German firms reported productivity gains from AI, suggesting it's already being used to streamline and consolidate workforces. When productivity rises without headcount growing, that's often a sign that AI is somewhat replacing workers. So concerns about job losses in these regions are understandable. Now in emerging markets like in Brazil, the dynamics is almost the opposite. Labor is abundant and relatively affordable. So there is less economic incentive to replace workers with AI. Instead, companies are more likely to use AI to complement their workforce, helping employees do more, reach new customers or access new markets. This expands output and can actually create jobs rather than eliminating them. There is also a structural factor. Emerging markets can leapfrog older technologies entirely rather than automating legacy systems, they can build new AI-enabled industries from scratch in fintech, health care or logistics, for example, generating fresh employment in the process. So in short, in developed markets, AI optimizes existing systems in high wage environment, creating some pressure on headcount, whereas in emerging markets, AI can enable growth in a labor-abundant environment, creating new opportunities.

Kristen Hallam:

All right. So that leads us to the question, do many emerging and frontier economies have enough skilled professionals to develop and manage AI systems? I'd like to bring you in to talk about that, Alisa. What does the talent pipeline look like in Sub-Saharan Africa, for instance?

Alisa Strobel: 

Building on Polly's point about job creation, when you look at Sub-Saharan Africa, that optimism makes sense. We have the youngest, fastest-growing workforce in the world. So AI should be a huge job creator here. But there's a paradox. We're building world-class data centers faster than we are developing the local talent to run them. For example, in Senegal, the government aims to train 90,000 data science professionals for the next 2 years, yet most universities still offer general tech degrees with limited AI focus. So, leaving Senegal ranked just 10th in a regional AI talent readiness analysis done by the Gates Foundation. In fact, across the wider continent, only South Africa and Kenya cracked the top 5, trailing far behind several North African countries. Thus, without local expertise, scaling AI solutions for local challenges remains a struggle. According to a recent government survey in Nigeria, 80% of companies reported last year that insufficient tech skills are actively hurting their operations. Kenya is in a very similar position. A recent survey shows 99% of organizations view AI skills as essential, but they also report that talent shortages are delaying projects or stalling innovation. The few AI graduates produced are quickly recruited by multinationals or emigrate creating a training for export pipeline, so to speak. And now consider operational roles like data center technicians, cloud architects, network engineers, these face even wider gaps. So you end up flying in expatriates that drive up the costs. However, some countries like Rwanda and Mauritius are pairing infrastructure with structured skills programs. In fact, Rwanda ranks #1 continent-wide in government AI readiness, partly because their national strategy invests heavily in capacity building alongside infrastructure. But right now, they remain rather the exception.

Kristen Hallam:  

All right. Thank you, Alisa. So, zooming in further on the Sub-Saharan African region, and as you just mentioned, data centers, we're seeing headlines about large data center investments across sub-Saharan Africa. But what's the reality on the ground? Are these investments the silver bullet for AI adoption? Or are there deeper constraints at play, Alisa?

Alisa Strobel:  

Thanks, Kristen. Yes, to answer your question directly, yes, there are absolutely deeper constraints at play. And while I can zoom in on Sub-Saharan Africa today, it's important to note upfront that emerging markets elsewhere are facing similar infrastructure bottlenecks that are constraining their digital growth. But looking at the reality on the ground here, the data center expansion we are seeing is real, and it's being driven by genuine structural demand -- we are seeing major projects like the IFC's USD 100 million investment in Raxio Group facilities across the continent, the UAE's AI for development initiative that was launched at the 2025 G20 Summit, and even solar-powered computing campuses in Botswana. So we can say the demand drivers are solid. We're also looking into solid GDP growth projected through the next 10 years, higher urbanization rates and deeper regional integration under the African Continental Free Trade area. But here's the critical insight from our latest research: The macroeconomic upside of these data centers depends far less on demand and far more on execution constraints. Let me walk you through 3 specific bottlenecks we are tracking. The first is power reliability. Data centers are incredibly energy intensive, and our analysis shows that Sub-Saharan African grid constraints are fundamentally misaligned with these requirements. In Nigeria, operators are often forced to rely on diesel generators because grid power is unreliable. Or let's take Kenya, even with plans for a geothermal-powered Azure cloud region, the broader grid still faces chronic instability. The result, operating costs for data center providers inflate by 30% up to 40% and productivity gains get suppressed. The second constraint is foreign exchange liquidity I'd like to highlight. Data center construction is highly import-intensive. Service, cooling systems, power equipment are all sourced externally most of the time. And during the build-out phase, this temporarily widens trade deficits in a country. African countries often lack foreign exchange reserves because their external revenue from primary commodity exports are volatile and insufficient to cover high import costs and external obligations, leading to persistent imbalances. So overall, many economies are then confronted with harder financing costs for these imports. And third is institutional execution capacity. We've seen this movie before with infrastructure projects in the region, projects get announced, capital flows in, but then they hit bottlenecks, thin anchor tenancy, permit delays, hidden public exposure tied to grid upgrades or government guarantees. So I think that kind of sums it up. Data centers are absolutely necessary for AI adoption, but the constraint isn't demand, it's delivery. And until we solve the most pronounced execution constraints, many of these facilities risk being unutilized becoming rather more expensive in place than engines of broader economic transformation.

Kristen Hallam:  

Alisa, building on those institutional bottlenecks you just mentioned, let's shift to the regulatory landscape. Across various regions, we're seeing a highly fragmented regulatory environment. At the same time, many governments are pushing for data localization. How do these twin challenges, the regulatory vacuum and the push for data sovereignty collide?

Alisa Strobel:  

Yes, this is an important point. A few countries like Rwanda and Mauritius have comprehensive AI strategies. Many others are often operating under outdated laws that just don't address AI-specific risks. Let me give you some context. Out of the 20 major markets we attracted early last year, only 7 in Sub-Saharan Africa had formal national AI strategies, albeit this is slowly changing. Nevertheless, this lack of regulation is colliding with severe political volatility during the '24, '25 election cycles in Ghana, Senegal, South Africa, for instance, we saw an explosion of AI generated deep fakes and disinformation. Concurrently, as political unrest intensified, governments resorted to blunt force Internet shutdowns often citing misinformation as justification. For businesses, this creates what I call a massive wait-and-see tax. If you are a multinational bank or a tech firm, you may hesitate to deploy customer-facing AI because the legal liability is completely undefined. Now this regulatory uncertainty feeds directly into the second issue, the data localization trap. In a bid to regain control, governments are increasingly demanding data sovereignty, mandating that citizen data stays with the national borders. The logic makes sense for privacy and security, but here's the macroeconomic reality. They lack the infrastructure to support it. Outside of South African and Kenya, most countries have maybe 1 Tier 3 data center, if any, so companies are forced into an impossible choice: violate local data laws by using foreign cloud providers or don't deploy AI at all because local compute capacity doesn't exist. But here, what's challenging really for AI development is that AI requires massive diverse data sets to function accurately. If every country builds a walled garden around its data, we suffer from data set fragmentation. And in this instance, you cannot train a robust pan-African AI model. And we see that the only way out of this bottleneck is regional integration. The African Continental Free Trade area digital trade protocol, for instance, adopted in February 2024 is somewhat of a game changer as it aims to create a consistent rule set for cross-border data flows, allowing a sort of hybrid setup where you have sensitive data staying in the country, but the heavy AI processing happens at regional hubs. So if we don't harmonize these regulations and allow data to flow securely across borders, we won't develop our own AI ecosystem. And this applies to other regions with similar uncertain regulatory conditions as well.

Kristen Hallam:  

Thank you, Alisa. Those are really valuable regional insights that help illustrate some of the points that we've been discussing today. So, let's return to outcomes. Beyond efficiency, where are firms actually seeing growth? Right now, efficiency seems to be the primary driver of AI adoption. However, our research found 22% of firms attributed actual income growth to AI in 2025 and 40% anticipate revenue increases this year in 2026. So, Polly, how are companies transitioning AI from a back-office cost-saving tool to a front office revenue generator? And what impact does that have on their employee composition?

Pollyanna de Lima:

That's a great question, and it really ties into what we've been discussing. So, case studies from the PMI special survey showed that a lot of firms are using AI mainly for efficiency by automating repetitive back-office tasks like data entry, scheduling and processing invoices. But what's interesting is that by freeing up time from these routine tasks, employees can potentially focus on higher value work, things like building client relationships, closing deals or coming up with new ideas. So, when we see that 22% of firms attributed actual income growth to AI in 2025 and 40% anticipate revenue increases in 2026, it seems like companies are starting to make that shift. They are perhaps moving AI from just being a cost-saving tool in the back office to actually driving revenue in the front office. Now how does that happen? Well, maybe it's because employees who used to spend hours on admin work now have the time to focus on customers, sales and innovation. AI can sometimes handle the busy work so people can connect with clients, solve problems and creatively grow the business. But as for employee composition, we are already seeing a shift. The PMI data show reduced demand for administrative and customer service staff, for example, and a greater need for workers that can manage AI initiatives, like Sophie mentioned, AI software engineers, operations, platform administration and also see greater hiring of strategic thinkers for roles like applied research and development. So companies appear to be looking for employees who can use AI tools to amplify their work rather than just doing manual tasks. In short, by automating the repetitive stuff, AI frees up employees to focus on revenue-generating activity. And that's probably why more firms are expecting revenue growth. And as this happens, the types of skills companies look for may shift towards more strategic roles.