Laurent - stock.adobe.com
Does AI represent the tech industry’s irresponsible era?
To be economically dependent on the rise of artificial intelligence is a risk that some appear prepared to take
When Nvidia CEO Jensen Huang told a company town hall meeting in November that “the only thing standing between America and recession is us”, the only thing surprising about his statement was that it was true.
Huang’s comments merely echoed an earlier observation by Nobel Prize-winning economist Paul Krugman in August that the US would be heading into a recession “if the economy weren’t being supported by a huge boom in AI [artificial intelligence]-related investment. And this danger remains: if the AI boom goes bust, the odds are high that the US economy will be plunged into a recession.”
To have the world’s largest economy arrive at such a parlous state of affairs where it is almost entirely dependent on the success of a technology that is still in its infancy implies a degree of irresponsibility that is almost untethered from reality.
The amounts of money being spent on AI are staggering. Gartner recently forecast worldwide spending on AI would hit $1.5tn in 2025 and surpass $2tn in 2026. Separately, JP Morgan Chase estimates that more than $5tn will be spent on global datacentre and AI infrastructure and related power supplies over the next five years.
When you consider those eye-watering sums, you could be forgiven for thinking that politicians, much of the tech industry and some of its more “charismatic” leaders are pursuing an “AI at all costs” strategy. Such a strategy seems epically reckless and irresponsible to the wider population, and to the planet, given the consequences of the enormous amounts of capital, energy and water that will be required to power the Ozymandian scale of their AI ambitions. It’s also a strategy that appears to be still searching for an objective.
What will AI deliver to CEOs, organisations and companies currently being swept up in the hype? What exactly is the tech industry selling with AI today, and should they be buying? Have we entered the tech industry’s irresponsible era?
It’s probably worth noting that we have been here before, albeit on a smaller scale, with cloud computing. In that context, it’s interesting to note the results of Kyndryl’s recent Cloud readiness report, which found that 70% of CEOs had arrived at their current cloud environment “by accident, rather than by design”.
Blame game
While CEOs might be pinning the blame on themselves in this report, who sold them the dream of the cloud, and who provided the technology for their sprawling, splintered architectures without a clear-cut end goal in mind? The tech industry. It’s like blaming drug users and absolving drug dealers of all culpability.
The issues are amplified with AI and the effects would be far more wide-reaching. As more and more people warn of an AI bubble based on hugely exuberant valuations and massive, potentially unsustainable, planned capital investments, how damaging could it be for the wider economy if the bubble bursts? Are we heading for “AImageddon”?
Jon Gill, head of TMT and Venture at global law firm Eversheds Sutherland, observes that concerns of a bubble persist and that AI “fear of missing out” (FOMO) is driving activity. “We remain still in the early days of the AI revolution,” he says. “There are some unprecedented numbers being invested in both the stock market and private capital.”
“There will clearly be an adjustment at some point,” says Gill. “Though based on the deals we’re advising on, there is some way to go yet – in particular, in the infrastructure for AI, where you see the tech majors spending huge amounts of capital expenditure. While developments like the UK’s Tech Prosperity Deal are welcome, recent market movements show that uncertainty will remain a feature of the AI sector – even as the strongest players deliver sustained financial performance. The global race to secure access to state-of-the-art compute continues to accelerate.”
In a recent article in The Guardian, former OpenAI researcher Steven Adler stated: “There are people who work at the frontier AI companies who earnestly believe there is a chance their company will contribute to the end of the world, or some slightly smaller but still terrible catastrophe.”
In that light, it’s probably not reassuring to hear that Yoshua Bengio, one of the godfathers of AI, notably remarked at TED2025 in April: “A sandwich has more regulation than AI.”
Economic consequences
In many ways, this is all of a piece with the tech industry’s habit, in the words of Ed Barrow, CEO and co-founder of Cloud Capital, “of celebrating the next big thing long before it understands the economic consequences”.
“We saw it with cloud: extraordinary innovation paired with a remarkable lack of financial discipline,” he says. “CIOs chased speed, vendors chased growth, and no-one stopped to ask whether organisations were designing for resilience and cost control or just accumulating complexity by accident.”
He describes what’s happening with AI as “eerily familiar”. “The narrative is grand and the capital deployment astonishing,” says Barrow. “But behind the hype, most organisations still don’t have a clear handle on what AI will really deliver for their business or what it will cost them to operate at scale.”
He shies away from labelling this as an “irresponsible era”, but agrees we are “in a period where enthusiasm is outpacing financial clarity. The industry is selling possibility. Customers need to buy outcomes. And between those two things sits a chasm that must be bridged with governance, transparency and discipline.”
Ross Teague, CEO at Nebula Global Services, agrees that the cloud “arrived by accident” for many enterprises because technology “was too often sold as a destination rather than a strategic enabler. There are clear parallels now emerging with AI,” he says. “Hype, unchecked investment and vendor-led pressure risk pushing organisations into fragmented, energy-intensive AI architectures without defined outcomes, much like the first wave of cloud adoption.”
From his perspective, it’s not a question of whether AI will deliver value, but whether it will “deliver the right value, sustainably and securely”. Too many providers are selling AI “as an inevitability rather than an intentional transformation”, says Teague. “Customers shouldn’t be buying ‘AI for AI’s sake’, they should be investing in measurable improvements in efficiency, customer experience, resilience and operational excellence.”
He warns that an AI bubble is only avoidable “if organisations resist exuberance and work with partners who prioritise outcomes over purely output and hype. If the industry enters an irresponsible era, it won’t be because AI failed; it will be because planning did,” says Teague.
Hype over design
Matthew McDermott, director of data governance at Access Partnership, warns of “the risk of treating AI as an end in itself rather than a means to solve real problems”. The cloud era revealed what happens when technology “is adopted on hype rather than design”, he adds. “With AI, the stakes are far higher: we’re not just buying software licenses, we’re building energy-hungry infrastructure that could reshape capital markets and national power grids.”
There is too much focus on “selling ‘AI capability’ without a clear articulation of value”, says McDermott. “More GPUs don’t inherently deliver growth, productivity or trust. If we continue at this pace, we risk a bubble where capital investment outstrips economic return, leaving companies, and potentially entire regions, servicing debt on infrastructure they can’t effectively use.”
The contrast with the telecoms crash of the late 1990s and early 2000s was that it had the benefit of leaving fibre in the ground that proved valuable when the business case was found, but “GPUs will quickly become obsolete, and if the bubble bursts, many datacentres will become crumbling temples to hubris”.
It’s probably worth noting that IBM CEO Arvind Krishna recently did a breakdown of the figures behind AI in an interview with the Decoder podcast. “It takes about $80bn to fill up a one-gigawatt datacentre,” he calculated. “If one company is going to commit 20-30 gigawatts, that’s $1.5tn of CapEx … You’ve got to use it all in five years because at that point, you’ve got to throw it away and refill it. Then, if I look at the total commits in the world in this space, in chasing AGI, it seems to be like 100 gigawatts with these announcements. That’s $8tn of CapEx. It’s my view that there’s no way you’re going to get a return on that because $8tn of CapEx means you need roughly $800bn of profit just to pay for the interest.”
Ben Gilbert, vice-president at 15gifts, agrees with IBM’s Krishna that “much of the technology being hyped today won’t deliver the returns being promised”. While he believes a market correction to the AI bubble is possible, “it’s unlikely to be a dramatic dot-com-style collapse”, he says. “Some AI projects will be cancelled, but the underlying potential of AI remains strong, and therefore its adoption will continue to grow in the long term, once businesses devise a more strategic mindset to implementing it.”
One problem could be that businesses may be misdirecting their efforts regarding AI. “Right now, most organisations are investing in AI to drive efficiency – automating workflows, reducing manual effort or enhancing customer support, for example,” says Gilbert. “But these benefits take years to materialise and are notoriously difficult to measure beyond time savings. When budgets tighten, projects with unclear or delayed ROI will be the first to fail.”
Purpose and value
At some point, it might be worth people asking why most organisations are investing in AI to drive efficiency and who recommended that as the best area for them to focus on.
Justin Megawarne, managing partner at Megaslice, doesn’t mince his words. “There’s nothing that should be pursued at the cost of purpose and value to real people,” he says. “That includes AI. The tech industry will sell any shiny new toy they think terrified customers will buy, because the tech industry is in love with itself, and has no idea what ‘human purpose’ really is.”
Megawarne’s view is that “customers shouldn’t buy anything that isn’t aligned to their purpose and doesn’t have an aligned downside if the tech provider screws it up”.
“If the customer loses, the tech provider should be prepared to lose in proportion,” he says. “But that is rarely ever the case.”
Jonathan Wright, chief product officer at GCX Managed Services, is adamant that “AI isn’t a silver bullet for the channel, and AI at all costs is not yet a feasible strategy. In the near term, I believe that AI will be best used to deliver incremental use case-specific gains rather than a sweeping transformation.”
At present, “a lot of the industry is selling potential, and that’s where leaders need to be careful”, he says. “Customers should carefully consider whether they really need it, and work with channel partners to look through a clear ROI lens.”
Wright believes AI adoption “should be a gradual, long-term endeavour, solving clear use cases rather than a rapid transformation”.
“Channel partners must remain as the expert technical advisers and act as practical guides to help ground their customers’ AI adoption in practical use cases rather than blanket transformation,” he says.
Be upfront
Claus Jepsen, chief technology officer at Unit4, says vendors need to “be upfront about what is actually possible” with AI. People should ask themselves a few basic questions: “Do I even need AI? What problem am I looking to solve? How good is my data? Are my employees ready for AI?”
He’s unequivocal, however that AI “will deliver huge benefits to organisations”, but the biggest challenge for CIOs is “the ability to shut out the noise and hype around the technology”, which is critical to identifying tangible use cases where AI will not just deliver cost savings through automation, but redefine the way business processes operate and create new ways of working.
McDermott agrees that people need to be clearer about what AI can deliver. “The real test isn’t who can build the biggest model; it’s who can show measurable impact for citizens, customers and the public sector,” he says.
McDermott argues that the channel “should be acting as a brake on hype, forcing clarity on outcomes, rather than acting as cheerleaders for ever-bigger deployments. AI could power the next productivity revolution. It could also become the most expensive sunk cost in tech history. The difference lies in discipline, not scale.”
