Most companies begin with one tool, then gradually expand their toolkit
En apenas dieciocho meses, la inteligencia artificial ha dejado de ser una curiosidad tecnológica para convertirse en una presencia cotidiana en el tejido productivo argentino: cuatro de cada diez pymes ya la utilizan, y la mitad restante contempla sumarse en los próximos dos años. La caída de las barreras de acceso —impulsada por plataformas gratuitas como ChatGPT— explica la velocidad del cambio, aunque la profundidad de esa transformación sigue siendo, por ahora, modesta. Lo que el dato revela no es una revolución consumada, sino el primer capítulo de una historia cuyo desenlace aún está por escribirse.
- El 85% de las pymes que hoy usan IA comenzaron a hacerlo en 2024 o 2025, una aceleración que refleja cómo la gratuidad de las herramientas generativas derrumbó de golpe años de resistencia.
- La adopción es profundamente desigual: el sector de software lidera con un 85% de penetración, mientras que textiles, madera y papel permanecen rezagados por falta de capital técnico y presupuesto.
- La mayoría de las empresas usa IA para tareas básicas —generación de texto y código— y la aplica sobre todo en marketing y administración, no en el corazón de sus procesos productivos.
- El 57% de las pymes que adoptaron IA no mide formalmente su impacto, lo que convierte las expectativas optimistas —77% espera ganancias de productividad en cinco años— en una apuesta todavía sin respaldo empírico sólido.
- El empleo, por ahora, resiste: el 91% de las empresas adoptantes no redujo personal, aunque sectores como metalurgia y software anticipan pérdidas de puestos cercanas al 20% a medida que la automatización madure.
En apenas año y medio, la inteligencia artificial pasó de ser un fenómeno marginal a instalarse en el centro de la conversación empresarial argentina. Una encuesta de la UTDT revela que el 41,6% de las pymes ya incorporó al menos una herramienta de IA, y que otro 46% planea hacerlo en los próximos dos años. El motor de este salto fue la irrupción de plataformas generativas gratuitas o de bajo costo: más del 85% de las empresas adoptantes comenzaron en 2024 o 2025, y antes de 2022 la práctica era prácticamente inexistente en el sector.
Sin embargo, la amplitud del fenómeno contrasta con su profundidad. Casi el 80% de las pymes que usan IA se limita a herramientas de generación de texto y código —ChatGPT y similares—, y la mayoría las aplica en funciones comerciales y administrativas antes que en sus procesos centrales de producción. La brecha sectorial es pronunciada: el 85% de las empresas de software y tecnología ya opera con IA, combinando aprendizaje automático, análisis de datos y toma de decisiones automatizada, mientras que industrias tradicionales como textiles o papel muestran tasas de adopción mínimas, frenadas por la escasez de talento especializado y presupuestos ajustados.
Las empresas suelen seguir un camino predecible: empiezan con una sola herramienta, ganan confianza y van ampliando su repertorio. Más de la mitad trabaja con uno o dos recursos simultáneamente; solo una minoría ha avanzado hacia aplicaciones más complejas. Y aunque el 77% espera que la IA impulse su productividad en cinco años y el 64% anticipa reducción de costos, el 57% todavía no tiene ningún sistema formal para medir si eso ya está ocurriendo.
En materia de empleo, el panorama actual es tranquilizador: el 91% de las adoptantes no redujo personal. Pero la calma podría ser transitoria: metalurgia y software proyectan pérdidas de alrededor del 20% de sus puestos a medida que la automatización se profundice. Lo que los datos dibujan, en definitiva, es un mercado en movimiento temprano —la IA ya cruzó el umbral de la conciencia empresarial argentina— pero cuya verdadera transformación todavía está pendiente.
In the span of just eighteen months, artificial intelligence has moved from the margins of Argentine business life to something approaching the mainstream. According to a survey conducted by UTDT, just over four in ten small and medium-sized enterprises—41.6 percent—have now incorporated at least one AI-powered tool into their operations. Another half are seriously considering doing so within the next two years. The speed of this shift is striking, but what's equally striking is how narrow the actual deployment remains.
The vast majority of these companies are using AI for the same basic tasks: generating text and writing code. Nearly 80 percent of firms using AI rely on these generative tools, platforms like ChatGPT and similar language models that became widely available and affordable—or free—starting in 2024. This explains much of the recent acceleration. When the barriers to entry collapse, when a small business can experiment with powerful technology without major capital investment, adoption spreads quickly. More than 85 percent of Argentine companies now using AI began doing so in 2024 or 2025. Before 2022, the practice was essentially nonexistent in the SME sector.
But adoption is wildly uneven across industries. The software and IT sector stands apart, with 85 percent of companies using AI—nearly double the overall average. These firms aren't just dabbling. They combine text generation with machine learning, automated decision-making, and sophisticated data analysis. For them, AI isn't a helpful assistant; it's woven into the core of how they produce. Traditional industries tell a different story. Textiles, wood products, paper manufacturing—these sectors show the lowest adoption rates, constrained by the nature of their work, limited access to specialized technical talent, and smaller budgets for technology investment.
When companies do adopt AI, they tend to follow a predictable path. Most start with a single tool, usually a text generator, then gradually expand their toolkit as they gain confidence and see potential applications. More than half of AI-using firms work with just one or two technologies. Only a smaller group has moved to three or more simultaneous tools. The progression reveals something about how businesses learn: they begin with the easiest entry point and only later venture into more complex territory like machine learning or workflow automation.
The applications themselves cluster heavily in commercial functions. Marketing and sales lead the way, with 54.7 percent of AI-using companies deploying the technology there. Administration and management come next at 41.5 percent, followed by production and services at 40 percent. Research and development is gaining ground—32.7 percent of firms report using AI for innovation and product development. But here again, the software sector skews the numbers dramatically. Exclude software companies, and AI use in core production processes drops to 28.3 percent across all other industries. For most traditional manufacturers, AI remains a tool for back-office work, not the factory floor.
The question of whether any of this actually improves business performance remains largely unanswered. Fifty-seven percent of companies using AI have not set up any formal system to measure its impact. Among those that have measured results, 24 percent report positive outcomes. Only 1.2 percent found no meaningful benefits. The most commonly cited gains are modest: 16.8 percent saw productivity improvements, 11.6 percent noted better innovation capacity, 10.4 percent achieved cost reductions. Yet expectations for the future are far more optimistic. Seventy-seven percent of companies believe AI will boost productivity over the next five years. Sixty-four percent expect cost savings. Nearly half anticipate revenue growth.
On employment, the picture so far is reassuring. Ninety-one percent of companies that have adopted AI report no workforce reductions. Only 7 percent have cut jobs, and another 1.8 percent think they might. This likely reflects the current reality: most AI use remains experimental and supplementary, focused on assistance rather than replacement. But looking ahead, some sectors see risk. Metalworking and software both anticipate job losses around 20 percent as automation deepens. The investment picture suggests momentum will continue. Sixty-eight percent of Argentine SMEs say they either currently invest in AI or plan to start. Software and IT companies are investing most aggressively and planning to spend more. Textiles shows the most resistance, with a third of firms saying they have no plans to invest at all.
What emerges from the data is a market in early motion. AI has crossed a threshold in Argentine business consciousness. It is no longer exotic or marginal. But the current wave of adoption is shallow—concentrated in simple tools, applied mostly to support functions, measured inconsistently if at all. The real question is what happens next: whether companies will move beyond experimentation toward genuine integration, whether the gains will materialize as expected, and whether the uneven adoption across sectors will widen or narrow over time.
Citas Notables
The adoption of AI is in its early stages and still in an experimental phase— UTDT survey authors
For software companies, AI is not a support tool but the core of the production process— UTDT survey
La Conversación del Hearth Otra perspectiva de la historia
Why did adoption accelerate so suddenly in 2024 and 2025? What changed?
The tools became free or very cheap. ChatGPT and similar platforms removed the financial barrier. Before that, AI was expensive and specialized. Now a small business can sign up and start experimenting immediately.
So it's not that companies suddenly understood AI better. It's that they could finally afford to try it.
Exactly. The survey calls it a "latent market." Businesses wanted to experiment but couldn't justify the cost. Once the cost vanished, adoption followed naturally.
But the software sector is adopting at 85 percent while textiles lag far behind. Is that just about money and expertise?
Partly. But it's also structural. Software companies can build their own solutions. Textile manufacturers depend on buying off-the-shelf tools. And their production processes—cutting, sewing, finishing—don't map neatly onto what current AI does well.
Most companies are using AI for text generation and basic coding. That seems... limited.
It is. But it's also how learning happens. You start with what's easiest and most obvious. The survey shows companies gradually add more sophisticated tools as they gain confidence. It's a progression, not a destination.
Half the companies using AI haven't measured whether it actually works. Doesn't that seem reckless?
It reflects the experimental phase they're in. They're trying things, seeing what sticks. Formal measurement comes later, once adoption becomes routine. Right now they're still learning what's possible.
What worries you most about this data?
That the gains might not materialize. Companies expect big productivity improvements in five years, but the current evidence is modest. If reality disappoints expectations, investment could slow. And the gap between software companies and traditional industries could widen into something structural.