- 22 hours ago
Productivity is measured as the ratio between production and resources used to obtain it, in volumes. AI brings higher capacity to process data, enables faster execution and progressively drives humanless operations. In a nutshell, productivity is greatly impacted. But in a more unstable and fast-moving environment, the question becomes not only “how much”, but also “what kind of value” do we want to create. Is productivity only about doing more or also with greater impact and resilience over time? This session explores how leaders can rethink productivity for a world where access to resources and stability are no longer a given.
Category
🤖
TechTranscript
00:01Hi, everyone. Hello. We're very happy to be here. I'm Margo, partner at Serena. We're a European VC firm, active
00:11since 2008. We invest from pre-seed to Series B, and we invest across deep tech, applied AI and impact.
00:19And I personally lead what we call corporate partnerships, meaning I deploy and manage custom funds that we built with
00:27some of our corporate partners.
00:29And I'm Xavier Jaravelle. I'm a professor of economics at the LSE in London and also chair of the French
00:37Council of Economic Analysis, which is a think tank of economists that's independent, but also linked to the prime minister's
00:44office.
00:45So we're very happy to be with you today to discuss productivity, AI, and challenges for leaders.
00:51Exactly. We wanted to share perspective because we come from different angles that we find kind of complementary.
00:57So Xavier, as an advisor to policymakers with a macro perspective, and I spend my daily life between startups and
01:04corporate thinking how they could better collaborate.
01:07And the current AI situation is really calling for how will it impact productivity tomorrow.
01:12So maybe the best way to start is to take stock of where we stand as an economy.
01:18Yeah. So it's a session about productivity.
01:21So what is productivity?
01:22What's the amount of economic value we create with a given amount of input, so labor, capital?
01:28And as you know, if we look at France, if we look at Europe, compared to the US, the picture
01:34isn't that rosy.
01:35So if you compare the rates of productivity growth over the past 30 years, we have been slower than the
01:43US, and that means we've lost about 40% of GDP per capita.
01:48So it would be about 40% richer if we had had the same rate of productivity growth here in
01:53Europe and in France compared to the US.
01:56So where does this come from?
01:58So probably you all think about the tech giants.
02:00Say, well, that's where the value is created.
02:03And that's true, but that's only about 20% of the overall gap.
02:09Most of the gap, two-thirds of the gap, are actually about IT, but not about the giants themselves, about
02:17the diffusion, the adoption of IT technologies in many sectors.
02:24For example, retail is one big example.
02:28So we've been adopting IT much slower.
02:31That's a drag on productivity, and that matters enormously for public debt, tax receipts, welfare, you name it.
02:39And the danger today, of course, is that the same story happens with generative AI.
02:45We have some champions, but clearly not at the level of the US, and adoption is also slower.
02:50So just one number to finish, if you look at intensity of usage of generative AI in someone's day of
02:59work, there's some studies that say in the US, 13% of ours have an intense use of gen AI.
03:06In Europe, in France, it's only around 7%.
03:10So it looks like this diffusion lag is happening again.
03:14And I think, Margot, you've seen some of this firsthand in your activities.
03:18Indeed, I think what is interesting is that we observe that at a micro level, and what you just said,
03:24we really reflect on it.
03:26Because take very simple metrics like cloud adoption or software equipment.
03:32European companies are systematically lagging behind if you were to compare with the US.
03:38Of course, at enterprise and large enterprise level, this is converging, although we are slower.
03:44But at SMEs, definitely, the gap is still quite large.
03:48Just taking one example, cloud adoption in the US, 75% of SMEs migrated to the cloud.
03:55In Europe, it's still below 50%.
03:59And actually, you were mentioning we have to make sure we do not repeat the story with AI.
04:05This can have a cost because migrating to the cloud or adopting software means that you make sure to have
04:15a data stack that is scalable, queryable.
04:18You make sure to have clean pipelines.
04:20You open your architecture and documented APIs.
04:24So today, when it comes to implementing AI, your stack is actually ready.
04:29This is one of the first principles.
04:31But also, what we observe is that actually AI is entering companies via software.
04:39One interesting survey that actually Salesforce ran last year on their US firm is that more than 85% of
04:47their clients actually do try and adopt their AI features.
04:52So they start their AI adoption via unlocking those features within their own software existing stack.
04:59So we're lagging behind, but we're catching up.
05:04What does it take not to miss the AI boat today?
05:09So it takes many things, right?
05:10It takes the new tech giants, but also it takes, as we said, fast diffusion.
05:16And so I think the first point I'd like to make is that you might think, you might notice that
05:21public debate, at least in Europe, especially in France, is quite dominated by this worry about job losses.
05:27If we adopt AI fast, we're going to disrupt the economy.
05:31Many people will lose their jobs.
05:33In fact, what you see in studies is quite different.
05:35The companies that adopt AI faster actually preserve jobs better than others, including low-skilled jobs.
05:42The main risk is actually not to be replaced by AI, but have someone else replace you, outcompete you, someone
05:49else who uses AI.
05:50So adopting too slowly, I think, is the key risk.
05:54And that's also what we see actually in public policy discussions, not just in public debates.
06:01There's a lot of effort put into creating new giants with some success.
06:07There's much less on diffusion of these AI technologies.
06:12So we do little in Europe and in France especially, while other countries do much more.
06:18So Singapore, South Korea, some countries also closer to us, like the UK, have much more ambitious policies to help
06:28diffuse the technologies.
06:29So it's a mix of many things.
06:32So it includes workforce training, demonstrators to demonstrate use cases, especially to small and medium-sized businesses that might not
06:41have the resources to invest in this and figure out what works best.
06:44There's obviously the questions of regulation with regulatory sandboxes to facilitate experimentation.
06:52There is a lot of subsidies in some of these countries.
06:55So these time-limited tax credits, let's say, if you train your workers for AI, if you purchase certain types
07:01of software, you're going to get part of these investments back in the form of a tax credit.
07:07And there's also the question of facilitating the links between small startups, large firms, which is some of what you
07:15do, Margot, in your daily activities.
07:18So I think what we need is to have some institutions, probably at the national level but also at the
07:24European level, that really lead this effort to put on an equal footing the new giants of tech agenda and
07:32the agenda of diffusion of generally of AI.
07:36South Korea, Singapore, they have these sort of one-stop shop that coordinates the diffusion policy, which we are currently
07:44missing in Europe.
07:45Of course, a lot of this is a private sector thing.
07:49It's not just a public sector.
07:50A public sector can try to help.
07:51It can also do it directly, by the way.
07:52Tax administration, education, there's also a lot directly to do there by adopting AI.
07:57But for the private sector, there's a recent series of economic studies that highlights the role of management practices.
08:06So it seems that Americans are just better at adopting new technologies, reorganizing workflows, organizations.
08:14So if you compare, for example, American multinationals that are in Europe to European firms of roughly the same size,
08:21so they face the same structural constraints, the same environment.
08:23But it turns out that still in these local offices of the American multinationals, there's much faster adoptions of IT
08:3130 years ago and today, gen AI.
08:34So perhaps, Margot, you want to say a bit more about that private sector management perspective?
08:39No, no, that's super interesting because that adoption question is definitely at the heart of what we try to do
08:46when we, of course, want to build champions.
08:47We want them to be adopted by the other side of the economy, the established one.
08:51So to reflect on what you said, adopting technology, in my perspective, is a muscle.
08:57So to train a muscle, you need discipline and you need methodology.
09:01Actually, discipline is about culture, definitely.
09:05And we see a very clear difference whenever the leadership, especially the CEO, is trained and has good knowledge and
09:12a very strong level of awareness about AI.
09:14This changing completely, the dynamic of conversation in the EXCO, but also the pace at which the roadmap is reviewed
09:21and implemented.
09:23And actually, if you do not have that DNA, you can import that.
09:27Thinking of some recent moves of non-technology companies that went to poach.
09:31Actually, the CTO that were our tech companies.
09:34Look at Allianz CTO.
09:35It's the ex-CTO of SAP.
09:37Super interesting move, I guess.
09:39I really believe.
09:40And methodology is actually having a proper framework from A to Z, which means you need to have a clearly
09:47identify your needs.
09:48You need to have a very strict make-or-buy rule.
09:51You need to have a proper effort to implement technology, to train people, then to track usage so that you
09:57can then measure the ROI of the technology you're trying to adopt.
10:00This whole operational framework is not always super well actually in place, but it is something that is quite easy
10:07to step up to.
10:09I want to pinpoint on that make-or-buy rule because I really believe at the very moment now with
10:15AI, we kind of think we can all make by ourselves.
10:19One number, actually, in the U.S., if you look at adoption of AI, in 2024, 53% of AI
10:28solutions were purchased by enterprise.
10:31In 2025, it was 76% of purchase solution, VA's, MED solution.
10:39So as they mature, they really know how to identify and buy the right solution.
10:45Very last comment upon that when it comes to adopting technology for corporate is that good reason not to do
10:51it now, which is sovereignty.
10:53I completely understand the point because it's definitely something we need to focus on, but I do believe that we
10:59do have giants and actually we do have European solution.
11:02So it's more a question of identifying them and being very, very sure about where you want to have a
11:08proper sovereign solution because actually most of our stack is already in the hands of a U.S.
11:15Right. So we need to put the diffusion agenda front and center.
11:18That's kind of one of our main messages.
11:20That's both private and public sector responsibilities.
11:22But we also don't want to just forget the champions, the European tech giants agenda.
11:29And I think there is a risk that we diminish our efforts there because we might get discouraged.
11:35We know we have all these structural constraints.
11:38Giants already exist elsewhere.
11:39We don't have deep capital markets and so on.
11:42But we also have these new potential giants.
11:44So we're in Paris.
11:45We have Mistral.
11:46We have Ami Labs.
11:48And I think one risk is that we, at least from a public policy perspective, that we decide to reduce
11:53our efforts because we have other constraints.
11:56We have mounting public debt in Europe in general, especially in France.
12:00And I think that would be a grave mistake.
12:02So I want to just warn against that risk.
12:05We have very powerful investment programs to accompany these companies and many others.
12:13One of them is called France 2030.
12:15It's a bit like the DARPA program in the U.S.
12:18It turns out it works very well.
12:19In fact, it works so well that they invested about $50 billion over 10 years.
12:24And the returns they created, effectively, more than offset the initial investment costs from the government.
12:29So that's an agenda to keep pushing.
12:33And I think, Margot, you agree with this and with that perspective of, in a way, optimism about the European
12:38potential.
12:39Completely.
12:39And we do support any public initiative that is promoting that, especially in public procurement.
12:45Having some kind of a European preference is definitely something we advocate for very strongly.
12:49And, yes, I mean, our job is to find and then support those future champions.
12:53We definitely have the talent.
12:54I mean, we had fun mapping, actually, sorry for being chauvinistic, but more than 4,500 French talent at the
13:03top 50 AI companies in the world.
13:05So we definitely have people that have the skill set and have the capacity to build those.
13:09They actually already exist.
13:11And it's a chicken and egg problem to adopt sovereign solution and not to have tech champions.
13:16So it's also about jumping in the pool and be willing to collaborate with them.
13:20I suggest, therefore, any, actually, procurement people in the room to track the Eurostack initiative, which is mapping all the
13:29sovereign solutions that we have today if we want to implement them.
13:34Maybe moving forward to conclusion, because we wanted to have this conversation about the future of productivity.
13:40I think that's one of our key messages.
13:44We need not to miss that boat, and it's about adopting technologies that are coming in.
13:49And, yes, there is a clear tension with sovereignty because it questions the resilience of that in the future.
13:56And we think here it's about adopting an hybrid mode and not conceding speed to sovereignty.
14:01We need to identify where we really need to have sovereign infrastructure, and we need to push for it.
14:07We already have solutions, so it's about adopting them.
14:10But we need to keep up with the pace and to really focus today on how to diffuse and adopt
14:17those solutions.
14:19Exactly.
14:19And since a lot of the productivity gains are actually brought by much more than just bringing in the technology
14:26but also reorganizing the company, all this organizational change,
14:29I think for many companies you can actually afford to have maybe a solution that's slightly less efficient, maybe a
14:36European champion instead of an American champion.
14:38But if you do the work of reorganizing the company, you can have most of the gains there.
14:44So I think we want to end with that message of pushing for the hybrid mode of adoption where you
14:49would have – sometimes you would rely on the U.S. giants but often on the European ones.
14:55And that agenda of diffusion is one that is important both for the private and the public sector.
15:00So we're very happy to have a chance to share these ideas with you.
15:04Obviously, quite short.
15:05There's much more to say.
15:06We're going to be around after.
15:08So if some of you want to stop by to chat, we'd be happy to do so.
15:12Thanks a lot.
15:13Thank you very much.
Comments