Gary Marcus questions AI spending levels implicit in 9% of GDP forecasts
The argument reframes the AI capex debate: because consumer payment rates are tiny and AI search cannibalises ad revenue, the capital behind current buildout levels rests almost entirely on enterprise adoption succeeding. Investors and product teams evaluating AI revenue models should weigh whether labour-replacement savings and productivity gains can actually close the gap. The comparison to consumer spending on food, energy and software gives a concrete ceiling framework rather than a vague bubble narrative.
A WSJ analysis cited by Gary Marcus argues investors should be skeptical that Americans will ever spend on AI as much as they do on food — roughly twice what the nation pays for all energy or all computers and software.
Only 3% of US households paid for an AI service as of March, according to Bank of America data, while 50% of US consumers already use AI search for free, per McKinsey.
McKinsey projects 20 to 50% of click-through ad traffic is at risk as AI search captures decisions earlier in the consumer journey.
Marcus's report contends enterprises, not consumers, must drive AI profitability, with Goldman Sachs equity research in May stating that successful enterprise AI adoption will drive the economics for the entire supply chain.
no Trevor Noren @trevornoren WSJ: "Is it plausible that Americans will spend as much of their income on this one technology as they do on food? Roughly twice what the nation pays for all forms of energy or all computers and software? Seven times what consumers spend on phone, streaming, and internet services combined? You should be skeptical. Even the most transformative inventions eventually run into the law of diminishing returns: each additional dollar a user spends yields less additional productivity (or enjoyment) than the last. That imposes a natural ceiling. The question, of course, is where that ceiling is. Whether or not you think 9% of GDP is right, you have to care, because this figure isn’t some fever dream: it is implicit in the dollars that investors and companies are committing right now." As I dissect in my report on "The AI Trade" ( sageroadresearch.com/collections/re… ), it's also important to recognize that consumers are unlikely to account for much of that "9% of GDP": "The vast majority of consumers don’t spend money on AI, with only 3% of US households paying for an AI service as of March, according to Bank of America data. We’re doubtful that payment rate will increase significantly anytime soon. That’s in large part because AI is primarily used by consumers as a search engine and for the entire history of the internet, search engines have been accessible for free. According to McKinsey, 50% of US consumers now use AI search. AI platforms are now taking 15 to 20% of informational query volume, according to DigitalApplied. If anything, consumer AI use is a threat to search-driven revenue. AI search reduces ad interaction, with 20 to 50% of click-through traffic at risk as AI search results “capture decisions earlier in the journey,” according to McKinsey projections. Enterprises, not consumers, are the great short- and long-term hope for AI profitability, or as Goldman Sachs’ equity research team summed up in May: “Successful enterprise AI adoption will drive the economics for the entire supply chain.” The enterprise capital required to subsidize the AI buildout will have to come from savings generated via labor replacement or growth driven by increased productivity or a combination of both. Significant and persistent hurdles remain. To again quote the report: "In December, we warned that market participants were overestimating GenAI’s strengths and trajectory of improvement and underestimating its persistent weaknesses, which would hold back enterprise adoption. We won’t retread all of the evidence we presented to support that conviction, but to sum up the basic concerns: Models remain error prone, which constrains utility. Their contextual understanding is limited and ivory-tower prognostications about job displacement to come neglect the contextual understanding required to execute even low-level tasks across sectors. Finally, most legacy companies will be unwilling or unprepared to tear down existing operations enough to mitigate these challenges and maximize AI’s potential." WSJ link: wsj.com/tech/ai/will-a… 🔗 View Quoted Tweet📊 6 ⚡
This report is categorized under AITOP's "Industry" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include Gary Marcus, AI investment, AI search, McKinsey.
Source attribution: curated from Gary Marcus (@GaryMarcus).