AI’s Economic Footprint: Unmasking Its True Value Creation

Unlock the Editor’s Digest for free

Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.

The frenetic buzz around Artificial Intelligence has begun to coalesce into a more sober, albeit still optimistic, assessment of its practical utility. For financial markets, the crucial pivot is no longer *if* AI can deliver, but *how much* value it genuinely provides, and over what timeframe. This nuanced understanding is now directly influencing capital allocation and valuation models across the tech spectrum, as companies and investors alike grapple with translating AI’s potential into tangible bottom-line impact.

One particular point of tension between AI’s boosters and detractors has been the disconnect between reported increases in coders’ output and the apparent lack of a corresponding boom in product or value creation. A new paper by MIT’s Mert Demirer and co-authors offers crucial insights, leaving both sides able to claim partial vindication while highlighting critical strategic implications for businesses and investors.

The study tracked software developers’ work before and after they adopted AI tools, measuring impact at several different levels. This included granular metrics like the amount of code written and the number of discrete files edited, progressing through to the number of projects or features worked on, and ultimately, actual releases of new software.

The findings reveal a pronounced “funnel effect” for AI-driven productivity. An explosive impact was observed at the top of this funnel, with coders creating or editing almost 300 per cent more files. However, that impressive boost was halved to 150 per cent by the time they reached the number of discrete pieces of work submitted for review. This in turn shrunk fivefold to a roughly 30 per cent uplift in the number of full software releases. This divergence between task-level efficiency and market-ready innovation presents a critical challenge for investors assessing the return on investment (ROI) from substantial AI capital expenditure. Companies touting internal productivity boosts may not see those translate directly into top-line growth or expanded market share if bottlenecks persist further down the development pipeline.

While a 30 per cent uplift in producing a company’s core product is undeniably significant, the study’s findings nonetheless demonstrate how perceptions and even some direct measures of AI’s impact on productivity can be far out of step with the value it ultimately adds. What feels like — and indeed measurably is — an explosive boost for a particular task often translates into a much more modest gain once that work has passed through all the human and organizational bottlenecks associated with reviewing and releasing production-grade work.

Moreover, when the researchers looked at whether AI-assisted increases in software production have led to increased consumption, they found little evidence of a corresponding market boom. The marked increase in mobile app releases over the past year has not been accompanied by any increase in downloads — with most of the new apps failing to capture even a modest audience. This signals a maturing digital marketplace where increased supply doesn’t automatically generate demand. It implies that AI-driven development, while accelerating creation, is not yet a panacea for market differentiation or consumer acquisition, raising questions about the sustainability of investment in broad app development without a clear value proposition.

Notably, the finding that productivity and value creation have been much weaker than some assumed landed at a time when Uber CEO Dara Khosrowshahi revealed the company had blown through its entire AI budget for 2026 in one quarter, and was planning to switch much of its AI use to lower-cost models, reserving frontier tools for special cases. Khosrowshahi’s candid admission underscores the immense, often underestimated, capital expenditure involved in harnessing frontier AI. His subsequent pivot towards lower-cost models, reserving premium tools for niche applications, is a powerful signal to the market. This isn’t merely about cost-cutting; it’s a strategic recalibration of AI investment, moving from an “acquire at all costs” mentality to a more discerning, ROI-driven approach that prioritizes efficiency and practical application.

Then came new research on AI use for legal work, which found that pairing cheap open-source AI agents with top-end models acting as sporadic “advisers” delivered better results at much lower cost. Further validating this shift, this paradigm challenges the notion that “more expensive equals better” in AI, potentially disrupting the market for high-cost proprietary AI solutions and bolstering the ecosystem for open-source alternatives. For enterprises, this means a critical assessment of their AI stack, balancing the perceived superiority of frontier models against the proven cost-effectiveness and flexibility of open-source solutions.

It would not be unreasonable for some market observers to interpret all of this as evidence that AI’s capacity to deliver genuine, widespread value has been vastly exaggerated, or at least that splurging on the latest models is often unnecessary. However, Demirer and his co-authors feel the more likely explanation is that current organisational structures and marketplaces are not yet set up to take full advantage of the real underlying gains. This view is strongly supported by the evidence from past technological revolutions, where the real jumps in productivity and significant job displacement came from new companies and processes, rather than incumbents simply grafting new technology on to existing workflows.

Consider the case of electricity in the late 19th and early 20th century. Initial productivity gains were modest where factories simply replaced giant steam engines with giant electric motors but left the rest of the machinery and layout unchanged. The true boom in industrial productivity arrived decades later when engineers fitted individual workstations with their own small electric motors, fundamentally re-engineering factory layouts and workflows. This historical lens offers a crucial perspective for investors. The initial phase of any transformative technology often sees incumbents attempting to graft new capabilities onto old structures, leading to suboptimal outcomes. True disruptive value, and the associated market shifts, materialize when new entrants or radically restructured incumbents design their operations *around* the new technology.

The fact that incumbent software and knowledge work companies are finding only modest productivity gains by incorporating AI into existing workflows and organisational structures, while usage, revenue and productivity explode at Anthropic and OpenAI — companies built around AI, with products written and reviewed by it — is perhaps early evidence of the same dynamic playing out here, only much faster. This divergence offers a compelling narrative for investors: a “greenfield” approach to AI integration likely yields superior returns compared to retrofitting existing, often rigid, corporate architectures. For the market, this implies that the real beneficiaries of the AI revolution might not be the legacy tech giants, but agile startups or established players willing to fundamentally re-engineer their business processes and organizational DNA.

I suspect both camps are correct. A lot of corporate AI use and spending today is inefficient, reflecting the early-stage friction of a transformative technology. But realised productivity gains are capturing the interaction of powerful new tools with poorly suited structures and processes. These frictions and bottlenecks will only ease over time, as organizational learning accelerates and new, AI-native business models emerge, unlocking the full economic potential that markets are so eagerly anticipating.

[email protected], @jburnmurdoch

**Market Impact**

The evolving narrative around AI productivity has significant implications for capital markets, driving a fundamental reassessment of investment strategies and corporate valuations. Investors are likely to scrutinize AI-related Capex more critically, demanding clear pathways to tangible ROI and measurable shifts in competitive advantage rather than just “AI adoption.” This could lead to a bifurcation in tech valuations: companies demonstrating adept integration of cost-effective AI solutions and fundamental workflow transformation may see their market caps expand, while those merely “sprinkling” AI on legacy processes could face headwinds. The market for frontier AI models may experience pricing pressure as enterprises, following Uber’s lead, prioritize hybrid and open-source solutions. Conversely, firms specializing in AI integration, workflow re-engineering, and strategic consulting stand to gain as businesses seek to optimize their AI investments. Ultimately, the market is signaling a shift from an initial “land grab” for AI capabilities to a more mature, efficiency-driven phase, where genuine productivity gains, rather than mere adoption, will dictate long-term success and shareholder value across the technology sector and beyond.