The Accenture Barometer on AI and the Divide that Europe Fails to Address

On June 30th, Accenture launched the AI Progress Barometer, a biannual observatory measuring the level of artificial intelligence preparedness of around three thousand of the world's largest companies, recalculated every six months on a scale from zero to one hundred. The headline everyone picked up is this: in the last six months, European companies gained 1.6 points in preparedness, compared to 1.1 in North America, so Europe is moving faster. Unfortunately, it starts much further behind and remains there: the average European score is 43.1, compared to 48.9 in North America, and growing faster from a lower starting point hardly reduces the gap. Reconstructing the snapshot from the previous semester, the gap decreased from about 6.3 to 5.8 points: real progress, but far from overtaking.

Up to this point, this is what everyone has written. Then, I took the exact same numbers that Accenture disseminated to illustrate European recovery and analyzed them by company size, following the revenue threshold that the Barometer uses to distinguish “larger” companies from “smaller” ones. European companies with revenues over 10 billion dollars are now just 2.1 points behind their North American counterparts of the same size, 47.4 compared to 49.5. European companies below this threshold, which are still categorized as large global enterprises in the Barometer, remain at 7.6 points behind their North American counterparts, 40.5 compared to 48.1. The gap between large and “smaller” European companies therefore represents 6.9 points, whereas the same gap in North America is only 1.4 points. The internal European divide is thus almost five times that of America, within a universe that includes only large enterprises: this imbalance, I find, is more urgent than the lack of overtaking the United States. The overall competitiveness of a continent also hinges on how far behind the less prepared companies lag, a detail that the general average reported in headlines fails to reveal.

Preparedness or Enthusiasm? What the Barometer Measures
It's worth understanding what this readiness index, which Accenture calls AI readiness, actually measures. It is not the degree of enthusiasm for AI, nor the number of slides dedicated to the topic in investor presentations, but how much an enterprise already possesses the necessary ingredients to make AI work at scale. The Barometer looks at four axes recalculated every six months: strategic direction, i.e., the declared plan regarding AI and responsible AI; technological foundations, i.e., cloud, cybersecurity, and data quality; people and skills, i.e., how much the workforce is being retrained and how involved leadership is in the process; and process reinvention, i.e., how much the company is genuinely redesigning the way it works instead of merely adding a virtual assistant to the existing process. The numbers combine two proprietary tools from Accenture: the AI Index, an external assessment of corporate capabilities, and the Pulse of Change, a survey of executives conducted three times a year.

At the level of individual countries, the most marked progress comes from France (+5 points, up to 43.1), United Kingdom (+4.8, up to 44.5), and Spain (+4.6, up to 39.9). On the sector front, ten out of eighteen monitored sectors show improvement, with the fastest runners being insurance (+8 points, up to 48.6), travel (+5.7, up to 46.7), and consumer goods (+5.2, up to 43.7). In the insurance sector, Accenture's release describes a concrete example worth noting: the simplest claims processes are automated from damage assessment to payment, while complex cases remain with a human agent. This type of redesign only works if there are clean data, system integration, and a trained workforce to manage the new process: it is a good example of what it really means to rethink a flow with AI, rather than merely pairing a chatbot with standard work.

And What About Italy? The figure does not appear in the international release, but Teodoro Lio, CEO of Accenture Italy, disclosed it during the news announcement in Italy: an improvement of 2.9 points in AI readiness, a pace faster than the European average. Lio added that the real challenge now is to bring these benefits to small and medium-sized enterprises, which remain the backbone of the Italian manufacturing system. This is the same point I focused on in June when discussing the specificity of our production fabric, explaining why the risk of a two-speed adoption is more concrete here than elsewhere: I wrote about it in AI Apocalypse: True or False?, if you want to catch up.

Here it’s important to keep two levels together. On one hand, the Barometer observes a universe of large enterprises: even the “smaller” segment of the sample, below 10 billion dollars in revenue, does not correspond to typical European or Italian SMEs but to companies that are still very large in international comparison. On the other hand, the fact that the Barometer records a gap of 6.9 points already within this universe of large companies is a clear signal: if the distance between those with more capital and those with less is so evident in readiness, the risk that a similar dynamic reproduces, amplified, in the world of SMEs is more than a theoretical hypothesis.

The Entry Cost Doesn't Scale with Revenue
It’s worth explaining why I consider this problem more urgent than the missed overtaking of the United States. The cost to seriously enter corporate AI—that infrastructure of clean data, skills, and redesigned processes that the Barometer measures, quite different from the free chatbot anyone can open in a browser tab—has a strongly fixed component, which does not grow proportionally with revenue. Organizing data, building integration pipelines, setting up governance, investing in security and training are tasks that require time, specialists, and management attention, and that can weigh relatively similarly on a two-billion-revenue company and a twenty-million one, despite having different projects.

Large companies have the capital, margins, and dedicated teams to absorb this initial cost and afford long experimentation cycles, even with uncertain outcomes. Smaller companies, which in Europe and even more so in Italy employ the majority of workers, compete under the same market rules but almost never have a dedicated data department or a budget for structured training, and margins remain too tight to afford a year of experimentation on a process that may not work immediately. When the Barometer mentions that France is recovering 5 points or that insurance is gaining 8, it is mainly capturing what is happening in the boards of directors of large companies in those sectors. The rest of the production fabric, composed of companies with fifty or a hundred employees, remains outside the frame: not because it doesn’t exist, but because it is not the subject of this measurement.

Those with tighter margins wait, and by waiting often accumulate a backlog that over time can translate into lower productivity, lower wages, and a reduced ability to retain talented workers, who leave for companies that can afford to pay them better. The Barometer does not directly measure these effects, but it shows the difference in starting conditions: those who already have the data, skills, and processes ready can run, while those who do not must build them first. It is a slow, almost tedious process to describe compared to the apocalyptic scenarios circulating elsewhere, but this is exactly how a technological gap risks becoming an economic gap, piece by piece, without any newspaper headline signaling it until the number is already big enough to make news. The Accenture Barometer already shows this discrepancy; it just requires reading beyond the headline chosen for the front page.