Companies worldwide are pouring billions into Artificial Intelligence, chasing a revolution in productivity. But a sobering new analysis from investment firm Apollo Global Management suggests the payoff hasn't arrived for most firms. Why?
The Trillion-Dollar Disconnect
A recent report
from Apollo's chief economist, Torsten Slok, has sent a ripple of concern through boardrooms and investor circles. The analysis highlights a stark divide in the American economy. While a handful of tech giants—the so-called 'Magnificent Seven' including Microsoft, Google, and Nvidia—have seen revenue per employee soar from around $550,000 to nearly $800,000 since 2022, the story for everyone else is dramatically different. For the other 493 companies in the S&P 500, that same metric has remained almost completely flat, hovering near $550,000, where it has been since 2019. The conclusion is blunt: outside of the tech sector that builds and sells AI, the technology has delivered 'zero financial payoff' for the broader corporate world so far.
A Tale of Two Economies
The gap isn't just in revenue per worker; it's also in profitability. Profit margins for the Magnificent Seven have expanded significantly since 2023, while margins for the rest of the market have stagnated at around 10%. This suggests that while Big Tech is successfully monetising the AI boom, other industries are primarily experiencing AI as a cost. According to the Apollo note, many mainstream companies are beginning to view their AI spending not as a driver of growth, but as a cost centre that is weighing on their margins. This is leading some to pull back, scrutinising every dollar spent on 'tokens'—the units used to measure and charge for AI model usage. The data reveals an intense concentration of spending, with the top 10% of firms, mostly venture-capital-backed, accounting for over 95% of expenses on major AI platforms.
Echoes of Past Revolutions
This lag between investment and productivity isn't entirely new. History shows that transformative technologies, from electricity to the internet, often take years or even decades to translate into broad economic gains. This is known as the 'productivity paradox'. A 2025 study from MIT found that AI adoption can initially lead to a temporary decline in productivity as companies grapple with implementation. Firms must do more than just buy software; they need to redesign workflows, retrain staff, overhaul data infrastructure, and change management practices. Only after these complementary investments and adjustments are made does the growth truly take off. The current situation suggests most companies are still in the difficult, and expensive, adjustment phase.
The Human Factor
The challenge is as much about people as it is about technology. For AI to create value, employees have to use it effectively and consistently. This requires more than just access; it requires trust, training, and a clear understanding of how the tools fit into their roles. Experts point out that the organisations seeing returns are those that treat AI as a workforce transformation project, not just an IT deployment. Furthermore, as Slok notes, current AI models don't understand crucial business concepts like accountability, ethical trade_offs, or trust—all of which are fundamental to driving long_term revenue and profits. This means a human_in_the_loop is still essential, which limits the immediate, large_scale productivity boosts many had hoped for.
What This Means for India Inc.
For Indian companies navigating the AI hype, the Apollo report serves as a crucial piece of context. It's a reminder that simply investing in the latest AI tools is not a strategy for success. The real gains will likely come to those who focus on strategic implementation. This involves identifying specific business problems AI can solve, redesigning processes around the technology, and investing heavily in employee skills and capability development. The lesson from the S&P 493 is clear: without a clear plan for how AI will generate revenue or create efficiencies, spending can quickly become a drain on resources. The focus should be less on the 'wow' factor of the technology and more on the 'how' of its application to drive measurable business outcomes.
