NextFin News - Apollo Global Management co-President John Zito says too much of the artificial intelligence spending boom is going to what he called “low IQ” tasks. Bloomberg reported the remark on June 10, as technology and infrastructure investors keep pouring capital into chips, data centers and software tools with faster-productivity promises but unclear payback periods.
The comment adds force to a debate that has been building for more than a year: whether AI budgets are going to business problems that can support multibillion-dollar price tags, or to polished demonstrations that do little for earnings. Zito’s view carries particular weight because he is one of Apollo’s senior dealmakers, not a sell-side strategist or a startup founder seeking funding.
Apollo, one of the largest alternative-asset managers in the world, is known for underwriting risk, structuring capital carefully and investing where lenders or investors are paid for taking uncertainty seriously. As co-president of Apollo Asset Management, Zito has long been linked to a cautious, credit-focused approach centered on cash flow, downside protection and financing structures rather than market narratives. His skepticism should be read as a capital-allocation argument, not as a claim that AI is losing relevance.
The point is straightforward: not every AI use case deserves the same level of spending. Enterprise buyers are being asked to fund pilots, integrations and compute-heavy deployments even as many applications remain narrow, repetitive and relatively easy to automate. If a company spends heavily to replace a simple workflow, summarize routine documents or automate a task that was already low-cost, the return on investment may be limited even if the technology works.
Zito’s blunt phrasing matches a blunt risk. AI infrastructure has become a capital-intensive race, while many of the revenue models tied to it are still experimental. The distance between experimentation and durable profit can be wide.
That makes the remark less a rejection of the AI trade than a warning about valuation discipline. Apollo has repeatedly indicated that it sees real financing opportunities in the AI buildout, including power generation, grid expansion, data-center financing and related private credit structures. Bloomberg reported in May that Zito said AI could help power private credit growth, arguing that long-dated loans to hyperscalers, power companies and infrastructure providers fit Apollo’s underwriting model.
There is no real contradiction in those positions. Apollo is not saying the AI buildout is overdone across the board. It is saying capital should go to assets with identifiable demand, contractual cash flows and long useful lives, rather than to the most fashionable software feature of the moment.
That also helps explain why the comments are not a consensus view across the investment industry. Many strategists and corporate executives argue that AI is still at an early stage of adoption and that the first wave of spending, even if messy, is necessary to build the systems needed for later productivity gains. The cautious camp argues that many generative-AI deployments have been overhyped, that enterprise adoption remains uneven and that companies are finding it harder than expected to turn model access into margin expansion. Zito’s remarks fall firmly into that second camp.
The market backdrop makes the warning more pointed. AI-linked capital expenditure has become one of the dominant themes in equities, credit and private markets, with investors rewarding companies tied to compute, networking equipment, power assets and the most credible software monetization paths. That enthusiasm has lifted semiconductor valuations and utility demand forecasts alike. It has also increased the risk of indiscriminate spending, especially when boards and management teams feel pressure to show they are taking part.
Zito’s criticism also touches on how companies choose projects. New technology spending often gets justified through visible, low-complexity tasks because they are easy to demonstrate and easy to sell internally. But the easiest projects to show are not always the ones with the biggest economic payoff. AI can help with customer support triage, document classification and workflow automation, yet those uses may deliver only modest savings if the labor being replaced was already cheap or intermittent. More defensible opportunities tend to sit in larger operating bottlenecks such as procurement, logistics, underwriting, energy management or code generation at scale, where the value is measurable and recurring.
There is a financing case behind the warning as well. The AI buildout is colliding with a higher-rate environment, tighter credit selection and a market less willing than in the zero-rate era to fund speculative promises. For public and private companies alike, projects with heavy upfront capex, long payback periods and uncertain demand will eventually have to show that the model works. Apollo stands to benefit if capital markets become more selective, because pricing and structuring that risk is central to its business.
The divide Zito is pointing to is between AI spending that changes economics and spending that mainly creates activity. The stronger investments are likely to be tied to power, infrastructure, proprietary data, regulatory barriers or clear operating savings. The weaker ones are more likely to be built around fashionable features that impress users without materially changing returns.
One detail matters: Bloomberg’s June 10 report attributed the “low IQ” line to John Zito, not Apollo chief economist Torsten Slok. That places the critique with a senior capital allocator inside Apollo, making the message about investment discipline more direct.
Explore more exclusive insights at nextfin.ai.
