Robots Are Everywhere. So Why Do So Many End Up Back in the Box?
High costs, failed pilots, and the brutal math of supply chains are killing automation dreams. For small companies, survival may depend on robots that are easier to teach than to own.
You’ve seen the scene: an orange arm moving with the precision of a violinist, cameras watching every weld like hawks, and dashboards that promise “real time.” The future looks installed in the factory. And yet, too often that future stays stuck in pilot mode. Or goes back in the box.
In 2025, industrial automation is not a promise—it’s a filter. It separates those who turn PowerPoints into productivity from those who collect Proofs of Concept like trading cards. Beyond the “robot yes/robot no” debate, the real pattern is not just technical. It’s about business, supply chains, and—more and more—software.
The uncomfortable number: the “pilot trap”
For years, the most repeated stat in manufacturing has been brutal: more than 70% of companies that invest in Industry 4.0—robots, advanced analytics, AI, or 3D printing—never make it past the pilot stage.
This isn’t urban legend; IndustryWeek, citing the World Economic Forum, reports it as a structural challenge: scaling something that works in a test area to a whole plant (or a whole network of plants) is where many demos die. Moving from shiny lab videos to the messy real world… is not easy.
Spain is no exception. The Spanish 2025 Industrial Digitalization Barometer shows 13% of companies have not automated anything at all, and another 22.5% are stuck at pilots. Translation: about one-third of Spain’s industrial fabric is not capturing real benefits from automation.
And it’s not just “try and learn.” An older article in Cinco Días (2017) reported that 36% of Spanish companies canceled digital transformation projects (many tied to automation) because of costs and lack of ROI. An old number, yes, but still painfully relevant.
Who controls the supply chain… controls the robots?
To understand why robots work brilliantly in some companies and crash in others, you need to look at their spot in the supply chain.
Think about building a car:
Tier 1 are suppliers that deliver directly to the car maker (the OEM). They produce big systems: a full dashboard, a complete seat, a brake module.
Tier 2 supply smaller parts or subassemblies to Tier 1. For example, the plastic pieces that go into a dashboard, or the components of a brake pump.
Tier 3 are further upstream. They handle raw materials or basic parts like metal profiles, screws, machined parts, or cut sheet metal.
Now, how do robots play here?
Tier 1: huge volumes, millions of identical parts. Paradise for robots—fast, repetitive welding, painting, assembly.
Tier 2: more variety. Not everything is identical, but there are repeating “families” of parts. Robots can still shine if combined with reconfigurable tooling and smart software.
Tier 3: chaos. Small orders, different parts every week, razor-thin margins. A fixed robot here becomes an expensive toy. What works better? Cobots, 3D vision, quick-change tooling, and especially software to reprogram without headaches.
The takeaway: the higher you are in the chain, the more product control and repetition you have, the easier and more profitable it is to automate. The lower you are, the more you need flexible, software-driven robots to avoid them becoming expensive statues.
I’ve seen startups spend 60% of their yearly budget on a shiny cobot, only to realize they also needed a specialized engineer (€45k/year) to keep it running. Three months later, the cobot was running at 30% because of tech incompatibilities. Six months later, the startup shut down. (Startups Españolas).
The flexibility headache
Take a metal workshop (calderería) or a small manufacturer. One month they work for Big Company A, next month for Big Company B. If they use robots, it’s hard to get good returns—every project needs reprogramming, which is neither cheap nor easy. It requires highly trained staff.
I once heard in a forum that 85% of Spanish companies are Tier 1, 2, or 3. Meaning: very few control the final product, manage production on their own terms, or do long production runs. And that’s exactly where robots deliver the biggest impact.
Some numbers prove it: in 2024, Spain installed 5,160 industrial robots. Almost half (44%) went to the auto sector. Then came metal (16.5%) and food & beverage (12%).
Years ago, a colleague from a big Japanese company told me that many Spanish clients ended up returning their advanced cutting and sheet-handling machines. Why? They couldn’t get real performance out of them, and programming new jobs took longer than expected. Robotization is hard—but maybe not robotizing is worse.
Today, only 7.8% of Spanish companies use robots (about 1 in 13). Among large companies, it’s closer to 1 in 5.
The numbers are telling: while the annual adoption rate of cobots in Europe is around 30%, the truth is that this figure is heavily skewed toward established companies or startups with big funding rounds.
For the average Spanish startup, with a life expectancy of 3.5 years and resources always stretched thin between product, talent, and marketing, the upfront investment of €50,000+ per cobot (McKinsey data) feels like a leap into the void without a safety net. (Startups Españolas).
The future of robotics is not hardware—it’s ease of use
Main obstacles today? High upfront cost of robots, lack of skilled staff to maintain them, and the difficulty of integrating robots into existing production processes.
Experts say many projects fail because of poor planning, not involving specialists from the start, or a mismatch between the tech and real business goals. Add employee resistance to change, and you see why so many initiatives collapse.
That’s why small companies are at risk. They’re the most vulnerable: without money or internal know-how, robots can crush them instead of help them.
This is why I believe the future is not just more robots—it’s robots that are easier to program and reprogram. Imagine teaching a robot its task with Meta’s AR glasses, or training it in a new job using NVIDIA’s virtual environments. That’s the direction where usability meets survival.
Where is the future going?
We’ll see.