Two years ago, the lack of a ‘Chief AI Officer’ position in the structure was considered market suicide. Today, its maintenance is. The festival of burning cash for fear of FOMO is over, and the virtual assistants created by generative AI now mean only one thing for companies: a bleeding P&L. The ruthless cleaning of balance sheets has begun.
Verification
Between 2023 and 2024, technology strategy was dictated by PR. Simply weaving the acronym‘AI‘ into the quarterly report was enough to drive up stock market valuations. Budgets were bursting at the seams with subscriptions, licences and ‘proof of concept’ projects to prove to shareholders that management had its finger on the pulse. The result? Thousands of implementations stopped at the stage of expensive, absolutely useless internal chatbots. We got software that hallucinates with impressive confidence about the company’s holiday policy.
Now the dust has settled. The supervisory boards are saying ‘check’.
It turns out that 90% of supposedly ground-breaking projects are mere overlays (wrappers) on third-party providers’ APIs. They have a new logo, a gigantic implementation cost, but zero barrier to entry for competitors. Boards of directors have created silos. Newly appointed CAIOs (Chief AI Officers) have built ‘Centres of Excellence’ completely disconnected from the core business. Instead of streamlining billing processes, they were carving out bots integrated with the company’s Slack that nobody uses. This position, set up hastily to appease analysts, is now losing its raison d’être. These corporate evangelists are leaving in silence after several months in office. No one wants to bask in the glow of admitting that they burned through CAPEX on an expensive toy.
The profit and loss account has no sense of humour
Let’s understand the mechanics of this market: the cloud giants and hardware manufacturers have long since claimed their pot. They sold the shovels. The boards have taken on all the risk of gold mining.
Inference costs are rising at an exponential rate. The costs of integrating models with legacy systems and corporate data lakes have punctured the worst estimates. OPEX is swelling and the anticipated automation of labour costs has proved a delusion. The LLM model has not replaced five juniors in analytics or legal. It has simply created the need to hire an expensive senior to verify that the machine has not made up the data in the contract. The return on capital (ROIC) from these investments is negative.
The market consensus assumed that generative AI would optimise everything. Wrong. In 2026, the winners are not corporations flexing their virtual advisers on homepages. The winners are those who are quietly implementing analytics into the dullest back-office processes. Route optimisation in logistics, dynamic pricing in e-commerce, predictive maintenance algorithms in factories, ruthless risk scoring. Therein lies the real margin. In the hard mathematics, not in the lines generated by an algorithm.
Hangover in Central and Eastern Europe
This global trend is hitting the CEE region with redoubled force. The Polish, Czech or Romanian market has been growing for decades on cost arbitrage, BPO/SSC services and robust engineering. In 2023/2024, the region tried to copy Silicon Valley by force. The problem is that there are no VC funds on the Vistula River ready to subsidise experiments for a decade, waiting for one “golden shot”. In our part of Europe, business has to spin in Excel from the first quarter.
Managements in our region allowed themselves to be blackmailed by the pressure of the trend. We burned through tight R&D budgets on ‘GenAI’ type implementations that the end customer did not want to buy. Software houses rebranded themselves as ‘AI agencies’ in a panic, trying to sell Western clients someone else’s APIs as their own IP. Western markets priced it out and rejected it. Poland is the logistics and manufacturing hub of Europe. Our Holy Grail is to reduce the cost of the last mile by fractions of a penny and cut power consumption in machine parks. Instead of hiring hard data engineers to optimise these processes, CEOs preferred to pay for visionary workshops. The pursuit of status and complexes has lost us.
The Bottom Line
Innovation implemented purely for PR purposes is a capital stock crime, and the game of technological pretence has just come to an end. Buying licences out of fear of being left behind (FOMO) is not a strategy.
fIf the flagship AI project doesn’t cut operating costs by 15% over the next two quarters, reduce the full-time cost base or create a new quantifiable revenue line – shut it down at 8am on Monday. Fire the visionaries. Get rid of the evangelists. Hand over the AI budget to a ruthless COO. He’s tasked with finding profits in old databases and supply chain optimisation. Maybe it’s time to stop drawing technological unicorns in front of investors and get back to the hard craft.
