In a research note analyzing fourth-quarter earnings, senior US economist Ronnie Walker noted that discussions surrounding AI completely overshadowed what was fundamentally a strong quarter, with core corporate earnings (excluding the energy sector) growing a solid 4.6% year-on-year. Amidst this market enthusiasm, Walker writes that “we still do not find a meaningful relationship between productivity and AI adoption at the economic level”. However, the data revealed an important sign of something bigger to come: a median reported productivity gain of around 30% for two specific, localized use cases.

Walker’s analysis adds some real meat to a debate that has rocked Wall Street — and many retail traders’ portfolios — as several doomsday viral essays about AI eating into the economy fall into actual stock-market volatility. AI executive Matt Schumer and the highest finance Substack, Study of Citrineboth warned that AI could be more capable of doing white-collar work, and sooner, than many people think. Top executives including Microsoft’s Mustafa Suleyman (“human-level performance of most, if not all professional tasks” will be automated), Amazon’s Andy Jassy (“you don’t need a lot of people”) and at JPMorgan Jamie Dimon (“now’s the time to start thinking about it”) added their voices to the chorus.

Torsten Slok, the influential chief economist of Apollo Global Management, wrote in his Daily Spark on Saturday that “the dramatic shift in recent weeks in the markets’ narrative from ‘the economy is strong’ to ‘we’re all out of work’ is truly remarkable.” He argues that markets are beginning to believe the view of “techno-optimists” about the productive capabilities of AI in agreement with the Federal Reserve and economists.

To a master-data-cruncher like Slok, it doesn’t make much sense that AI expectations are “causing a macro conversation about the future increase in the unemployment rate,” because he sees no change in the “quiet future economic story of a strong US economy driven by AI spending, the industrial renaissance and the One Big Beautiful Bill.” Slok added that he thinks this narrative is wrong, that AI adoption will take longer than the next 12 to 18 months discussed in these viral essays, and that the risk of an overheated economy is greater than, say, unemployment up to 10%.

Goldman agrees with Slok at least that the vibes are a bit fearful, titling its report “AI-anxiety,” and highlighting how chat companies are far more than tangible implementations. A record 70% of S&P 500 management teams discussed AI in their quarterly calls, with 54% specifically framing the technology around productivity and efficiency. However, when it comes to providing hard numbers, the narrative falls short, supporting the research of Wharton management professor Peter Cappelli, who has many companies trying to adopt AI and said before luck that productivity gains are real, but getting there is hard work and relatively expensive to implement.

Only 10% of S&P 500 management teams have actually quantified the impact of AI on specific use cases, Walker wrote, and only 1% have quantified its impact on revenue. Moreover, broader economic adoption remains slow. While half of the companies in the broader Russell 3000 are discussing AI, US Census survey data shows that less than 20% of establishments are currently using AI for any business functions.

Here comes the “but.”

But AI has a big impact in 2 areas

Despite the lack of a macro impact on the entire economy, companies that have successfully integrated and scaled AI have reported significant improvements. Goldman Sachs found that management teams that quantified AI-enabled productivity impacts on specific tasks experienced median gains of nearly 30%.

Two main areas are driving these many gains:

• Customer support
• Software development tasks

In these targeted functions, technology is already delivering transformative promises, greatly facilitating business operations.

It’s probably not wrong, then, that doomsday predictions come from tech types who have seen firsthand how 30% of software development work has been lost to the future development of robots. Venture capital billionaire Marc Andreessen famously predicted more than a decade ago that software would “eat the world,” but software has found itself depleted. Goldman offered some clues as to how big the appetite for AI might be from here.

The earnings data suggest to Goldman that local productivity gains are beginning to influence corporate hiring strategies, leading to a “nascent reluctance to hire in anticipation of potential productivity gains”.

Walker observes a modest but increasing share of management teams that clearly mention AI when discussing hiring freezes or layoffs. Companies that discussed AI in the context of their workforce reduced their job openings by 12% last year, a higher decline than the 8% decline seen across all companies. While the current correlation between AI adoption and broad labor market outcomes remains small and statistically insignificant, Goldman’s baseline forecast is that 6% to 7% of workers – roughly 11 million jobs – will eventually lose to AI automation in the long term.

Even without widespread productivity gains, AI is dramatically changing capital spending. “Hyperscalers” — the big tech companies that provide cloud infrastructure and AI — are driving an unprecedented growth in spending. Analysts revised their expected capex in 2026 for these technology giants to a staggering $667 billion, a 24% increase from just the start of the earnings season and representing a 62% jump compared to 2025. Goldman Sachs expects this AI spending to contribute roughly 1.5 percentage points to the overall measurement. which is 1.5 percentage points of growth this year, although a slight increase in GDP in total of 1. to 0.2 percentage points due to the great reliance on imported capital goods.

Ultimately, Goldman’s findings paint a picture of an economy in transition. While Wall Street is consumed by “AI-anxiety” and tech giants pour hundreds of billions into infrastructure, the promised productivity revolution remains localized to software coders and customer service representatives. For the broader US economy, the true macroeconomic benefits of the AI ​​revolution have yet to arrive.