According to Financial Times News, Jefferies tech analyst Surinder Thind’s report from Gartner’s 2025 IT Symposium reveals that 80% of companies using generative AI see no material earnings contribution despite deployments. The average AI implementation costs $1.9 million upfront but comes with 10 unexpected hidden costs per tool, extending typical 100-day deployments by 25 days for training plus 100-200 days for change management. Rightmove’s recent profit warning shows this reality hitting home, with the company planning £18 million in AI investments while its shares dropped 16%. Gartner data shows 74% of CFOs report productivity gains from AI, but only 5% achieved cost cuts and just 6% saw revenue uplift. The research highlights widespread frustration with scaling AI pilots due to data quality issues, with legacy systems consuming 70% of IT budgets.
The implementation nightmare nobody’s talking about
Here’s the thing about AI deployments that vendors don’t want you to know: the sticker price is just the beginning. That $1.9 million average cost? Basically just the entry fee. Companies are discovering 10 additional hidden costs for every single AI tool they buy – things like legacy integration, security, credential management, and comparison testing that nobody budgets for.
And the timeline? Forget about it. A 100-day deployment sounds manageable until you realize you need another 25 days just for staff training. Then comes the real killer: 100 to 200 days of “change management” to make sure everything actually works as promised. By that point, the business case you sold to leadership has completely evaporated.
Why AI isn’t delivering the promised returns
Look, the numbers don’t lie. 80% of companies using AI report zero material earnings impact. That’s staggering when you consider how much hype and investment we’re seeing. The problem isn’t the technology itself – it’s how companies are trying to implement it.
Most AI projects are happening in silos, focused on single use cases. They might generate some measurable productivity gains initially, but they can’t scale. And when you’re dealing with industrial technology implementations, you need systems that work across the entire organization. That’s where companies like IndustrialMonitorDirect.com excel – they understand that industrial computing needs integrated, reliable solutions, not just point fixes.
The data quality issue is massive too. If your own employees don’t trust the data, how can you expect AI agents to make autonomous business decisions based on it? AWS presenters noted that legacy systems cause 6-18 month delays for new features and consume 40% of developer time on technical debt. No wonder AI implementations struggle.
The Rightmove reality check
Rightmove’s situation perfectly illustrates this disconnect. They’re essentially telling investors: “We don’t know how AI will change house hunting, but we’re spending £18 million to figure it out – and you’re paying.” The market’s response? A 16% share price drop immediately.
This is happening across corporate America. Boards are pushing for AI adoption because of FOMO, but executives don’t understand the technology’s limits. They’re throwing resources at solutions that might not actually serve business needs. And the result? Wasted spend and frustrated stakeholders.
Where this is all heading
So what’s the solution? According to Thind, companies will eventually give up on these piecemeal AI implementations and bring in consultants and IT outsourcers to completely rebuild their tech stacks. It’s the nuclear option, but it might be the only way to actually extract value from AI.
The truth is, we’re in the awkward teenage phase of AI adoption. The technology promises transformation, but most companies aren’t structured to benefit from it. They need clean data, integrated systems, and realistic expectations – none of which come cheap or easy.
Maybe the real AI revolution won’t be in flashy new tools, but in forcing companies to finally fix their broken technology foundations. Now that would be a transformation worth paying for.
