Is AI killing the tech industry?

The rise of artificial intelligence has sparked a huge debate: is AI destroying the tech industry or simply reshaping it from top to bottom? The feeling of vertigo is real: careers are being called into question, business models are faltering, and there's an avalanche of tools that seems unstoppable. But if we quiet the noise a bit, what we find isn't so much a technological apocalypse as a change in the rules of the game.

Today, almost anyone can access advanced AI models through a web browser, but this democratization doesn't mean all organizations are effectively leveraging these capabilities . AI is straining hardware demand, transforming the job market, forcing a rethink of data governance, and testing the true digital maturity of businesses. The key question is no longer whether we use AI, but how we integrate it, who governs it, and what we are willing to delegate to it.

Is AI really harming the tech industry?

One of the first visible consequences of this wave is the impact on hardware: AI models consume a massive amount of computing power and memory , leading to a race to develop specific chips and processors. This pressure has diverted resources from the manufacture of more traditional components, creating bottlenecks and increasing the cost of some basic consumer technology.

This shift in investment towards specialized AI infrastructure has made many previously relatively affordable technology products more expensive or harder to obtain . For end users and small businesses, this translates into slower equipment upgrades and a sense that the market is focusing on AI at the expense of other everyday technological needs.

In the workplace, the disruption is most noticeable in entry-level positions: many junior tasks in development, support, and content creation are being partially or fully automated . This can discourage new professionals who don't see a clear fit if they don't specialize in data, AI, or automation from the outset. The industry risks dangerously depleting the pool of generalist talent that typically sustains long-term technological evolution.

However, entirely new fields are opening up at the same time. New roles are emerging in areas such as data engineering, MLOps, algorithmic governance, model auditing, AI ethics, and the design of customized solutions. Companies that understand this shift don't see AI as an enemy, but rather as an engine for reshaping their value proposition , combining automation with high-value-added services.

An example of this approach is that of companies developing custom AI-powered software to optimize processes, improve customer experience, or strengthen cybersecurity . These organizations integrate advanced machine learning techniques with consulting services, cloud migration (AWS, Azure), and business intelligence (Power BI and similar tools), focusing the technology on generating better decisions, not just cost reductions.

AI and businesses: between discourse and reality

Looking at SMEs and professionals , the data is clear: there's far more talk about AI than actual day-to-day use . Recent reports on the Spanish business landscape show that around 45% of companies admit to never using AI in their operations, while only about 11% claim to use it regularly.

Between these extremes, 36% report using AI occasionally for specific tasks , reflecting a scenario of isolated tests, one-off pilots, and little structural integration. Most of these companies consider themselves "fairly digitized": almost eight out of ten say they have a medium or high level of digital maturity, but when you look at the specific tools they use, the picture changes.

In practice, only around 14% report having dedicated AI software integrated into their systems . This means that while basic digitization (ERP, e-invoicing, cloud-based office applications) has progressed, the adoption of AI and intelligent automation remains sluggish. The technology is no longer science fiction, but it's still not a standard part of many organizations' operations.

This gap between rhetoric and reality shows that the challenge is no longer access to technology, but knowing what to use it for and how to integrate it into real-world processes . Companies that simply "add AI" as a marketing ploy, without clear use cases or value metrics, end up generating internal noise and frustration rather than competitive advantages.

At the same time, industry studies indicate that AI is increasingly perceived as a matter of national competitiveness, not just individual company competitiveness . Executives and government officials agree that its impact spans productivity, employment, education, and institutional quality, and that a shared strategy is needed to prevent it from becoming a collection of isolated experiments.

The biggest barrier: education, culture, and time

When companies are asked why they aren't making the leap, the majority answer is clear: the main barrier to implementing AI is a lack of knowledge and training . Around 37% of organizations cite the lack of specific skills as the number one obstacle to progress.

This deficit is compounded by other very specific obstacles: 26% perceive AI as too complex a technology from a technical standpoint ; 25% say they don't see clear use cases that justify the investment. In other words, it's not so much that the technology isn't available, but rather that many decision-makers don't know how to implement it in their businesses.

In addition to these doubts, day-to-day operations work against transformation . Nearly 24% mention that they simply don't have time to implement new tools, 22% believe the cost could be high, and around 21% point to resistance to change within their own organization or among their clients. Inertia is a significant factor, especially in small and medium-sized enterprises.

All of this paints a picture where the barrier is much more cultural and organizational than purely technological . Most companies have heard of AI and recognize its potential, but lack the basic practical literacy needed to safely experiment, measure results, and scale what works.

This skills gap isn't limited to technical profiles: only a very small percentage of companies have implemented comprehensive AI training programs for their entire workforce . More than 20% don't have any structured program at all, and the training that does exist is concentrated in IT or data teams, leaving out managers and middle managers who are precisely the ones prioritizing investments and redesigning processes.

Too many hours spent on manual tasks that AI could alleviate

Paradoxically, while companies hesitate to adopt AI, they continue to dedicate an enormous amount of time to manual and repetitive tasks . Recent studies, for example, estimate that around 26,3 hours per month are spent on manual invoicing, about 32,3 hours on accounting, and about 24,5 hours on human resources management.

It is precisely these routine tasks where automation and AI could deliver a significant leap in productivity : automatic data extraction, accounting reconciliations, document classification, report generation, or top-notch customer service through well-trained chatbots. However, the adoption of these solutions remains limited and fragmented.

The paradox is that, when asked about the benefits they expect from digitalization, companies clearly identify the advantages : reducing time spent on administrative tasks, decreasing errors and costs, and even improving customer service. They know the theory; the problem lies in turning those aspirations into concrete projects, with clear responsibilities and follow-up.

Here, AI is beginning to create a divide between companies: those that use it only to "save time" and gain a little more efficiency, and those that combine it with a redesign of how they make decisions . In the latter, algorithms not only automate processes but also provide structured information to make better, faster, and more contextualized decisions.

In this environment, professional value shifts from "producing" to "discerning ." With tools capable of generating dozens of text, image, or analytical proposals in seconds, human work becomes focused on selecting what is worthwhile, what fits the strategy, and what risks it entails. Thinking isn't replaced; the point in the process where it becomes essential is simply changed.

The role of consultancies and law firms in the age of AI

Consulting firms and professional offices occupy a key position in this whole puzzle, especially for SMEs and the self-employed. Currently, most companies still see their advisor as a mere administrative manager , focused on tax and bureaucratic obligations rather than supporting their digital transformation.

Data suggests that approximately 49% of companies perceive their advisor as simply a tax and paperwork handler , while only around 5% consider them a true "digital partner" capable of guiding them in the use of new technologies, including AI. The opportunity for growth is enormous.

Interestingly, the consulting firms themselves acknowledge that they are not yet fully prepared for this change in role . Nearly 35% admit they need more training in digital and automation tools to effectively lead their clients' digital transformation. In other words, they too are stuck in manual tasks that they could be delegating to intelligent systems.

If these firms rely on AI for automatic document classification, draft report generation , or predictive analysis of accounting data, they can free up time to offer a much more strategic service , based on proactive advice, financial planning, or early risk detection.

In this new scenario, the true value of consulting will lie not in filling out forms, but in interpreting data, explaining scenarios, and helping clients decide what to automate, what to keep in human hands, and how to govern the systems they adopt . AI thus becomes a lever for professional repositioning, not a direct competitor.

What AI do we have today: specific, generative, and far from science fiction

It's important to clarify what we mean when we talk about AI, because not everything circulating in the public debate corresponds to the technology that is actually available . The vast majority of current systems are "narrow" or specific AI: models designed for concrete tasks, without general understanding or awareness of any kind.

Within this category, we find machine learning, deep learning, and generative AI , capable of producing text, images, audio, or code from large volumes of data. All of this is already deployed in commercial products and everyday services, from virtual assistants to recommendations on various platforms.

In contrast, concepts like AI general (AGI), capable of reasoning autonomously and flexibly like a human , remain, to this day, a theoretical horizon and object of research, but not an operational reality. Confusing these two levels fuels exaggerated fears and commercial hype that hinders sound decision-making.

The key point is that the AI ​​innovation process is already fully underway in almost every sector and country . The pace is so rapid that the main discussion has shifted from technical (what can be done) to organizational and social: how is all of this governed, who assumes responsibility, and what criteria are used for its deployment in businesses, government, and everyday life?

On a personal level, generative AI is becoming a ubiquitous productivity tool : drafting, summarizing documents, translating, suggesting ideas, and supporting learning. These functions will be integrated almost seamlessly into applications and devices, raising a new form of "digital illiteracy": not understanding how they work and ending up as passive users of systems whose logic we don't comprehend.

Real risks: errors, biases, and loss of control

The rapid deployment of AI brings with it very concrete and already visible risks: incorrect information presented with complete certainty, biases inherited from training data, a lack of transparency in some models, and serious impacts on employment and privacy . These are not futuristic problems, but issues of the present.

These risks largely stem not so much from the technology itself as from the lack of clear criteria for how it is designed, adopted, and monitored . Without control structures, AI tends to amplify the quality—or lack thereof—of the organizations that use it: where there are clear processes and robust leadership, it multiplies capabilities; where chaos reigns, it accelerates errors and conflicts.

In this context, it is important to distinguish between regulation and governance . Overly detailed or premature regulation can ultimately stifle research and innovation, especially in a global environment where other countries operate with more flexible frameworks. The example of the European AI Act illustrates part of this dilemma: its complex risk-level classification and prescriptive obligations have been criticized for being late and insufficient in the face of a rapidly evolving landscape.

At the same time, operating without any regulatory framework or common standards is not a realistic option either . The complete absence of rules fosters opaque practices, irresponsible data use, and the deployment of systems that can cause significant harm before anyone is held accountable.

What is beginning to gain traction in the international debate is a hybrid approach: governance based on self-responsibility, technical standards, and external oversight, supported by regulations that demand transparency and accountability without dictating every detail of how technology should be developed . The competition is no longer just about having the best models, but about making your standards a global benchmark.

AI governance models: from the enterprise to global agreements

In this area of ​​governance, several operating models can already be distinguished. The first is the corporate model of internal governance , adopted by large technology companies and data-intensive organizations: explicit ethical principles are defined, internal review committees are created, impact assessments are carried out, and models are implemented before being launched into production.

In this model, responsibility lies with the organization itself , which assumes the legal, economic, and reputational consequences if its AI systems cause harm or infringe on rights. It is a flexible model compatible with rapid innovation, but it depends on appropriate incentives and the existence of external oversight mechanisms to prevent it from becoming mere marketing.

A second approach relies on governance based on technical and professional standards , driven by international organizations and industry consortia. Standards are developed on data quality, robustness, security, traceability, and risk management. Although often voluntary, their adoption becomes a de facto requirement for operating in certain markets when mandated by clients or regulatory bodies.

In parallel, a multi-stakeholder governance model is gaining traction , in which businesses, governments, the scientific community, and civil society work together to reach consensus on principles and best practices. Initiatives such as the Partnership on AI and the Global Partnership on Artificial Intelligence are examples of these spaces where efforts are made to align diverse interests and lend international legitimacy to certain criteria for responsible use.

Finally, there is the regulated governance model , which does not replace the previous ones, but rather provides them with coherence and coverage. Here, the priority is not to regulate technology in detail, but to require organizations to have clear, documented, and verifiable governance models : who decides what, with what data, with what controls, and who is responsible when something goes wrong.

Training and responsibility: the piece that is often missing

Beyond committees and regulations, the expansion of AI makes it clear that the greatest challenge is human: how do we train people to use these tools without losing their judgment or autonomy ? Governance cannot be limited to systems and organizations; it must include the end users themselves.

AI literacy is not just about knowing how to program models: it involves understanding their limitations, model drift and biases, when to trust an answer and when to question it . In an environment saturated with automatically generated content, this critical thinking ability becomes essential to avoid falling into a blind dependence on what the screen says.

Effective governance, therefore, requires shared responsibility for its use, supported by ongoing training . Companies need their teams—technical, managerial, and operational—to understand enough to integrate AI into their processes without delegating everything to the algorithm. Citizens require educational tools to avoid confusing plausibility with reality.

Today, many organizations are operating with AI without clear rules of engagement : around one in four have no internal policies on responsible use, almost 40% do not follow standards or certifications, and a significant fraction admit that they do not adequately manage data privacy in their AI projects. This combination of improvisation and lack of knowledge is what truly puts both the technology industry and public trust at risk.

Ultimately, the question of whether AI is "killing" the tech industry falls short: rather than destroying it, AI is forcing it to reinvent itself, to rethink its business model, its relationship with talent, and its rules of trust . The difference between winning and losing in this context will not be who has the most brilliant algorithm, but who is able to combine technology, training, governance, and accountability to transform AI into a sustainable competitive advantage, and not a passing fad that generates more noise than value.