AI's ROI reality check: high spending, hardly any returns.

Executives have never been more all-in on artificial intelligence - yet many are asking why they aren’t seeing the payoff. Global enterprise AI spending is projected to reach $2.59 trillion in 2026, a staggering 47% leap over last year ([1]). Nearly four in five senior leaders worldwide now rank AI as a top investment priority ([2]), and AI is moving into everyday work at 22% of organizations ([3]). But momentum isn’t the same as success.

Multiple new studies underscore just how widespread the ROI shortfall is. Domino Data Lab’s latest Enterprise AI survey found 57% of companies still don’t achieve returns above the cost of their AI investments ([4]). An MIT Media Lab analysis of 300 firms was even more sobering: 95% of corporate AI projects show no measurable impact on the bottom line, with only 5% delivering meaningful financial benefits ([5]). And a global CEO poll by PwC revealed 56% of chief executives saw neither cost reductions nor revenue growth from AI in the past 12 months, while only 12% saw both ([6]).

In short, record investment hasn’t yet paid off for most organizations. These sobering findings are no longer just an IT concern – they’ve become a boardroom issue, forcing leaders to justify escalating AI budgets in the absence of clear returns.

One reason ROI is lagging is that the hidden costs of AI have exploded, ushering in a new era of financial scrutiny. Over the past two years, frenzied generative AI adoption - spurred by usage-based pricing models - led to unplanned budget blowouts.

The poster child is Uber. After giving 5,000 developers access to an AI coding assistant, the ride-hailing company blew through its entire 2026 AI budget in just four months ([1]). In a May earnings call, Uber’s COO admitted the AI spend had become hard to justify, noting the link between those costs and real business benefit was 'not yet there' ([2]). Even AI’s biggest backers are hitting limits: Microsoft reportedly canceled some AI tool licenses to rein in runaway cloud bills ([3]), and one unnamed enterprise discovered it had incurred a $500 million AI bill in a single month by failing to impose usage caps ([4]).

These financial wake-up calls are ending the era of blank-check AI projects. A new survey of 260 finance executives finds 60% are spending more on AI than they can justify ([5]), and only 22% of finance leaders can link AI expenditures to business outcomes today (even though 87% say they’ll need to within a year) ([6]). Recognizing this, vendors have begun offering tools to help tie costs to value: for example, IBM recently launched Apptio AI Value & ROI, a platform that shows how AI’s cloud usage (like costly model token calls) connects to business results ([7]).

Investors have taken notice, too. This week, Alibaba’s stock plunged nearly 10% after the Chinese tech giant announced a $10 billion share sale to fund new AI projects ([8]) – just days after revealing a 75% drop in quarterly profit due to heavy AI spending ([9]). The market’s verdict was clear: even in a tech boom, shareholders are punishing big AI bets that lack clear, near-term returns.

What’s causing this yawning gap between investment and impact? New evidence suggests the problem isn’t the algorithms, but how businesses deploy them. Companies often treat AI as a plug-and-play add-on, layering intelligent systems onto broken processes. A recent McKinsey study found that while three-quarters of organizations have incorporated AI into their operations, more than 80% report no discernible improvement in profitability so far ([1]). The biggest differentiator was process design: the few firms seeing profit boosts were far more likely to have fundamentally overhauled core workflows to fully leverage AI’s capabilities ([2]). Yet only 21% of companies in the survey had reworked even some key processes to maximize AI’s impact ([3]).

Compounding the issue is a lack of clear metrics and accountability. Fewer than one in five organizations track key performance indicators to measure AI’s business outcomes ([4]), leaving most firms flying blind as to whether their efforts are paying off. Many initiatives scale up with no defined value target, reminiscent of the last decade’s digital transformation wave when new systems were celebrated but often failed to deliver bottom-line gains ([5]).

Culture and talent factors also drag down ROI. One global survey found 54% of C-suite executives say rapid AI adoption is actually causing internal turmoil, and 60% plan to part ways with employees who refuse to adapt to AI-driven changes ([6]). With stakes this high, it’s little wonder so many AI projects stall.

Despite the grim statistics, a small group of companies is managing to extract real value from AI – and their playbooks offer lessons. Boston Consulting Group labels these organizations 'future-built' AI leaders (roughly the top 5%) and finds they achieve five times the revenue gains and triple the cost reductions of their peers ([1]). Notably, what sets these standouts apart isn’t spending more on fancy algorithms, but a relentless focus on tying each AI initiative to a concrete business outcome.

These leaders start with well-defined problems and measure everything. Walmart, for example, used AI to update 850 million product data points, greatly improving inventory and pricing decisions ([2]). DoorDash’s AI voice assistant for phone orders saved millions by automating customer calls ([3]). And Bank of America’s virtual assistant Erica now handles millions of client interactions, cutting call-center volumes by nearly 50% ([4]). In each case, the company targeted a specific use case – from supply chain optimization to customer service – and tracked tangible metrics like cost savings, revenue lift, or customer satisfaction improvements ([5]) ([6]).

Another common trait of high-ROI AI programs is discipline in cost and governance. Leading firms set clear “guardrails” on AI spend: they budget by use case, monitor cloud consumption closely, and pull back projects that don’t deliver. They treat AI not as a magical black box but as a business transformation enabler – which often requires reengineering processes and upskilling employees to fully capture value. Crucially, executive oversight is strong: companies where top leadership directly oversees AI strategy and ROI are dramatically more likely to see meaningful returns (e.g. 57% vs 21% reporting value, according to KPMG) ([7]). The bottom line for senior leaders: Realizing AI’s promise isn’t about chasing hype or spending freely, but about pairing technology with process change, cost control, and clear metrics. The few organizations that get this balance right are already leaving their competitors behind.