If you look at the current moment through the lens proposed by Oxford University professor Carl Benedikt Frey in his book How Progress Ends, it is difficult to avoid the feeling of déjà vu. Its pages challenge the conventional wisdom that economic and technological progress is inevitable. In fact, for most of human history, stagnation has been the norm, and in reviewing why some societies prospered and others failed over the last 1,000 years in the wake of rapid technological change, Frey sees a recurring tension in history: while decentralisation encourages the exploration of new technologies, bureaucracy is crucial to scaling them up. And when institutions fail to adapt to change, stagnation, if not outright regression, inevitably follows.
We are witnessing a contradictory dynamic today. Technological and scientific advances increasingly demand a capacity for collaboration, both horizontally, between different sectors, disciplines and disciplines, and horizontally, between different sectors, disciplines and disciplines. stakeholders; The value chain is a vertical one, between all the components of the value chain. From large corporations to individual people sharing their talents, through the companies of the middle market and SMEs, society is made up of a variety of nodes of specialisation that can mutually enrich each other. However, the institutional environment, especially in the wake of the identity crisis in Western countries, promotes fragmentation, polarisation and isolation. And if institutions do not help to scale up innovation breakthroughs, the risk of general sluggishness increases, according to Frey's vision.
There is, however, a substantial difference with respect to previous stages, one might object to his argument: the leaders of the technological industry who drive the generation and application of knowledge have such a power of initiative in the transformation of the economy and society, so spur and host scientific-technological research and control the flow of data in so many areas and planes of reality, that the main hallmark of the New Order we are heading towards, if the current dynamics persist, will probably not be stagnation, but the growing decoupling, at the strategic level, between the public and private sectors. The former, de-globalised, with the resource of force still in its hands, but a growing debt; the latter, globalised, dominated by capital, but subject to the law where it is operative to apply it.
This INTEC 2025 Report is intended as a roadmap for those who want to transform society with an inclusive vision, for the benefit of citizens in this time of turmoil. The alternative is social Darwinism in its multiple meanings, be it capitalist, statist or the product of de-institutionalisation. This reality will necessarily prevail, because the transformative effect of the new cycle of artificial intelligence (AI) and of advances in new materials, healthcare, geospatial technologies or smart grids, all of which are the subject of this work, can only come about if we act with an ecosystem vision, integrating all the actors in the value chain. Wherever they are.
The resolution of this challenge will have a direct impact on people's individual lives. Philippe Aghion, professor at the College de France and the London School of Economics, predicts that productivity growth thanks to the new AI cycle will be between the 1.3 points of the electric revolution and the 0.8 points of the digital revolution over the next ten years. As a result, a new intangible economy will be consolidated, characterised by uneven, constrained and not necessarily balanced TFP (total factor productivity) growth. In such circumstances, competition between territories to make the most of the new technology will be fierce. Only those able to adapt their institutions to this new intangible-intensive economy will emerge stronger. Those that learn to promote integrated strategies in which all public-private actors collaborate.
Nobel laureate Daron Acemoglu has calculated that, at current costs, it would be cost-effective in the US to automate 23% of tasks involving the visual factor. An MIT experiment on the impact of ChatGPT estimates that it can produce cost savings of 27%. However, these productivity gains are not being passed on to wages. The OECD has confirmed the slowdown in average wage growth relative to aggregate productivity, a phenomenon known as productivity-wage decoupling.
The labour transition in which we are immersed could be described as historic when viewed in perspective. A small set of companies, which it calls «superstars», have a dominant position in the market. This status increasingly protects them from competition, thanks to a combination of high productivity, high margins and a workforce polarised between high and low labour incomes. The OECD warns of productivity divergence: productivity is growing in technology frontier sectors and stagnating or barely advancing elsewhere. Robotisation is reducing the relative wages of workers in the middle deciles of the occupational wage distribution and the content of job tasks is becoming less fixed, as is the allocation of time to each task. The World Economic Forum (WEF)[1] It detects a rapid and growing transition towards business services jobs (the most common in Spain), in positions such as business analyst and sales representative, and towards digital jobs, such as software developer, generally at the expense of lower-level technical occupations. What is relevant is that, in the Forum's view, what is characteristic of this era is that job transitions no longer occur «naturally».
Mercer Marsh Benefits and the Reward and Benefits Association (REBA) agree that, to rebalance the market, it will be essential to develop people and talent with the right skills for the new intangible economy. That is probably the most important risk for employers right now: 23% of jobs will have changed by 2027, according to the WEF, with 69 million new jobs and 83 million existing jobs displaced. All this, in a context of massive labour underutilisation, with a global employment gap already exceeding 11% and with approximately 20% of young people (aged 15-24) not in employment, education or training, according to the International Labour Organisation (ILO).
Even recent tech graduates are running into difficulties and now account for just 7% of hires at large companies.[2]. While they are hiring 25% fewer young people than in 2023, start-ups are doing even worse, with only 6% of their new hires being recent graduates. The backdrop of political instability and intense rivalry over AI, coupled with shrinking venture capital, is causing Series A tech start-ups to be 20% smaller than they were in 2020, and this is leading them to opt for smaller and more experienced teams, accentuating the trend.
44% of the skills required in a given job will be transformed over a five-year period. Strategic workforce and talent development planning that incorporates faster and more efficient job reallocation mechanisms, within and across firms and sectors, is desirable. Public leaders and managers will have a responsibility to help workers move out of declining occupations and into the growing jobs of the future. New systems of skills accreditation, beyond traditional academic qualifications, are proposed. Credentials that could be based on a new standardised taxonomy of skills that professionals could use to move between roles, sectors and regions. Governments must also think about policies that balance labour flexibility with worker protections and benefits. By 2030, 92 million people could be working entirely remotely.
The ability to build collaborative networks will be key to addressing the challenges of new employment in a way that strengthens social cohesion while boosting productivity. Such cooperation needs to involve multiple stakeholders (multistakeholders), whether public or private, educational or productive, and will be facilitated by transferable benefits, which move with the worker from one job to another.
The deployment of AI takes the task of cohesive society through employment to a new dimension. Two years after the launch of ChatGPT, the number of job offers asking for AI skills had doubled. One of the first questions raised by the new cycle of generative technologies is precisely how to determine the advantages of being a person versus a machine. Every industry, bank, construction company, institution, hospital, university, NGO, retail chain and even transport service is going to have to decide which jobs to assign to people and which to artificial systems, and it must do so with an urgency never seen before.
At the 2025 Davos Forum, Salesforce chairman and CEO Marc Benioff proclaimed: “This is a moment none of us will ever forget. We are going to be the last CEOs to manage only humans as a workforce”. Just a few days earlier, Nvidia founder and CEO Jensen Huang had announced at CES in Las Vegas the arrival of the “ChatGPT moment of robotics”. It will take three main forms, according to his vision: AI agents, which he calls information workers, able to move freely, learn, analyse and make decisions on their own; autonomous cars, to which this report devotes a chapter; and humanoid robots. Huang believes that the IT (information technology) areas of every company will be turned into human resources of AI agents in the future. Physical AI will be able to understand the language of the world, from physical dynamics to concepts such as gravity, friction and inertia, geometric and spatial relationships, cause and effect.
Unexpectedly, the human element is the key link in successfully extending AI, but the division of roles between people and machines is not going to be solved by a team of Silicon Valley technologists. Each organisation will need to think about what it wants to do, what outcome it is looking for and how it can benefit from the technology stack to do so by delivering a better quality product. The new business imperative is to figure out what we need to train for, the new digital divide will not be whether we have a device or not, but what we do with it.
As the technology is employed, a new sense of trust and transparency will also have to be fostered, because AI will interact on our behalf, but independently of us. Microsoft expects 95% of its code to be automatically generated by 2030, a similar percentage to that of other companies such as Google and IBM and identified by consultancies such as McKinsey and Accenture for the technology sector as a whole. As IT systems become more sophisticated, the trend towards valuing deep specialists in multiple specialties will grow. A recent research paper focuses, in fact, on the burgeoning figure of the ‘generalist expert’.’[3].
New AI-based systems work particularly well where decisions need to be made accurately, agilely and in real time. They are optimal for dynamic situations where action is constantly needed. At an early stage, the way to improve generative AI was to scale its training by gathering as much data as possible. The next wave, however, will have its greatest impact on industrial and other process-based activities, because research has been directed towards creating smaller and more specific AI models that are more efficient and can be trained with almost any kind of information modality, not just text, images and sounds. In this sense, we speak of the ‘marginal returns to intelligence’, in a sense similar to what economists use to refer to the marginal returns to capital, labour or land.
Again, it is not conceivable to embark on this path with a fragmented view. The fastest way forward for generative AI is to pool data from different sectors and even share it between competing companies. In the future, we should be able to ask a question and have a set of AI models work together to solve it. Changing the rules of the automation game will transform a wide range of business models: from software as a service, to energy as a service, to mobility as a service. The new philosophy is already pushing companies to join forces in collaborative platforms that converge in an immersive environment, designed by hybrid digital and physical computing, where extended reality (XR) could be part of the constellation, alongside metaverse, blockchain, quantum algorithms and AI. Until now, each sphere of this ecosystem was observed in isolation, without being able to compose the overall picture. Generative AI was the missing piece.
The ability to collaborate will mark the boundary between winners and losers in this new context. Secure ecosystems that ensure the integrity and value of information, autonomous supply chains that process billions of predictions per day, adaptive, where it is difficult to distinguish between the activity and the activity that is not being processed, will be key. online and physics, where the parts are interconnected, without friction. This means reimagining the entire system, something that has never been done before on this scale. The question is not so much whether AI will change the business world, but who will be able to take advantage of it first and best. In these circumstances, companies must avoid splitting their workforces into digital and analogue workers, and will need to ensure that they educate consumers to recognise the possibilities of AI implementation in the future.
Data therefore takes on a new value. The global amount of data is expected to grow from 33 ZB in 2018 to 175 ZB by the end of 2025, according to the Data Age 2025 and the Global Big Data Analytics Data Guide, published by IDC. More than 20% of these will be transformed into new data assets, i.e. relevant factors of production for economic and social development. Columbia Business School professor Laura Veldkamp argues that data is “digitised information” and that its true value lies in reducing uncertainty around a prediction, the playing field where AI shows its full potential. Hence the connection between new technologies and information management. Ultimately, according to Veldkamp, all risk is “a tax on the economy”. His boldest proposal emerges by taking the syllogism a step further: “all gains from data collection should be passed on to consumers in the form of lower prices”.
Boston Consulting Group believes that sharing data with competitors may intimidate industry executives, even though the biggest challenges facing the economy “will not be solved by one company working alone and using only its proprietary data”. The consultancy estimates the value of the data sharing opportunity at 2.5% of global GDP. Complex problems such as fraud detection or supply chain optimisation can be addressed more effectively through collaboration, sharing data across multiple players. This is being done in the US by auto insurers through the LexisNexis CLUE Auto platform; and in Europe by Airbus in the digital ecosystem. Skywise. Joint AI models for business ecosystems, economic sectors or government domains will increasingly be trained by a single trusted organisation, responsible for collecting data from each enterprise, without leaving their premises, and for setting up a federated learning model.
Siemens is promoting such an initiative around its industrial LLM and has already convinced 60 companies to share their data in exchange for future access to the resulting AI model. The shared platform MELLODDY has also been developed by a European consortium of ten pharmaceutical companies with the aim of accelerating drug discovery. The Data Labs, promoted by the European Union as integral components of the future AI Factories (one of which will be installed at the Barcelona Supercomputing Center), are intended to foster the provision, sharing and secure exchange of information. These are also the main objectives of the European Common Data Spaces, although for now there is no way to overcome fragmentation and both programmes operate as distinct ecosystems, each with their own governance rules, strategic objectives, business models and technical requirements.
The need for collaboration in information management is not confined to AI. In the retail sector, more and more companies are supporting orchestration platforms that allow for dynamic switching between multiple payment providers in real time. As digital identity is introduced, which the European Union wants to extend to 100% of citizens by 2030, adaptive authentication solutions will be more likely to be applied. The shopping experience can then be personalised according to risk profiles, forcing retailers to integrate with the most appropriate mobile wallet solution to build customer trust.
In the case of cities, the proliferation of technological tools to capture the multimedia data generated in them opens up an era of new opportunities for technological innovators and public managers. The concept of computational urbanisation is gaining prominence as a new paradigm based on the big data geography and AI. It will help to understand complicated problems related to urban dynamics, energy use, traffic patterns and environmental impacts. Harnessing the computing, caching and communication capabilities of connected vehicles will also complement the telecommunication network of cities in the future. It will enable many location-based services and increase the intelligence of the urban environment, applying complex network principles and game theory approaches. The European Union is promoting the Citiverse idea, with intelligent virtual agents behaving in a way that is compatible with the digital avatars of human users. At a later stage, the digital twins should be interconnected in a European CitiVerse, within the framework of Society 5.0, but the condition for this is that 100% of EU citizens have a digital identity by 2030.
Given the possibilities opened up by collaboration through new technologies, the dynamics of fragmentation seem to respond to an unhelpful asynchrony. It is paradoxical, for example, that the technology sector is currently debating between the possibility of a US-sponsored internet, which is almost completely free to join, and the one promoted by China, which requires prior approval. The novelty is no longer that countries such as Russia, Turkey and even Brazil are considering switching to the Chinese model, but that the West is analysing the possibility of requiring licences to have an active website, as in the case of radio, as France is proposing. In its complex drift, the battle over tariffs has led to the emergence of the tool online TARIFF, which allows import fees to be imposed on Python packages, a very common programming language.
One of the unintended effects of this contradiction between globalist and divisive forces stressing society is that, by mid-2025, the AI bubble was already bigger than the dotcom tech bubble of the late 1990s, according to Apollo. Against this backdrop of uncontained technology expansion, conventional AI-centric data centres with electricity consumption equivalent to 100,000 homes are conceivable. And significantly larger complexes are being built today, each requiring 20 times more energy (half the power of the city of Madrid).[4], and it is only the beginning. It is illusory to conceive of an expansion of AI processing infrastructures without adapting the power grid, generation systems and the spaces where they will be deployed, as discussed in one of the chapters of this report. By the end of the current decade, the US will use more electricity for data centres than for the production of aluminium, steel, cement, chemicals and all other energy-intensive goods combined.
The roadmap in times of uncertainty is collaboration. The INTEC 2025 report deals in depth with the development potential of 10 key technological areas, in which Spain can still position itself with its own voice, because they are at an emerging point and the barriers to entry are not yet exclusive. In drafting the texts, we have applied the principle of setting out the opportunities offered by concerted action. In the case of AI to improve health systems, the World Health Organisation (WHO) estimates a global shortfall of 10 million medical workers by 2030, although other projections increase this figure to 15 million. Global healthcare spending is already equivalent to the entire GDP of the European Union and by 2025 the sector will move more than 10 zettabytes, or 10 trillion gigabytes, of data. So yes, the mainstreaming of smart healthcare should be imperative. The reality of antibiotic discovery research deserves a separate chapter, especially considering the fact that no new family of antimicrobials has been registered since the 1980s.
Autonomous cars are already driving in cities in the US and China, although they are still subject to strict regulatory constraints: they must drive in prescribed zones, in which they have been extensively trained. Europe is holding firm on this, even though Waymo's autonomous vehicles have 78% fewer injury-causing accidents than the human driver and 81% fewer airbag accidents. In the case of biosensors, collaboration is required across a wide range of industries to make continuous monitoring a reality, from those innovating in materials such as microelectronics and batteries. The availability of geospatial data will impact the development of agriculture, resource efficiency and energy management, among many other applications. Biotechnology will ensure that food and raw material supply is not a point of potential vulnerability that makes societies less secure.
The area of alternative technological possibilities to current ones opens up a very interesting field for innovation. In the report we have chosen areas such as new abundant materials capable of producing effects similar to those of critical raw materials; nuclear energy options without uranium, with thorium as a realistic option on the table; and quantum algorithms, which allow processes to be optimised with the efficiency of quantum computing on conventional computers.
Innovation only unfolds its full transformative potential when it is used as a factor of cohesion. It has been shown that the dynamics of fragmentation, especially those promoted by the new forms of populism that are spreading throughout Europe, reduce the innovative activity of the territories.[5] and condition the collaboration capacity of its scientific-technological fabric. The transition in employment and the new ways of operating introduced by AI make it necessary to integrate the different areas of organisations, build ecosystems with the rest of the actors in the value chain, share data with both the public and private sectors and agree on strategies. In this report we identify areas of opportunity for innovators, key to a roadmap.
[1] The Future of Jobs Report 2025, WEF, 7 January 2025
[2] The SignalFire State of Tech Talent Report - 2025, SignalFire, 20 May 2025
[3] Unmesh Joshi, Gitanjali Venkatraman, Martin Fowler, “Expert Generalists”, martinfowler.com, 2 July 2025.
[4] Energy and AI, IEA, April 2025
[5] Andrés Rodríguez-Pose, Zhuoying You, Peter Teirlinck, The political extremes and innovation: How support for extreme parties shapes overall and green scientific research and technological innovation in Europe, Research Policy, November 2025, doi.org/10.1016/j.respol.2025.105307