--- title: "Wharton Paper Warns: AI Must Boost Productivity by 2.7x to Avert Tech Receivership" type: "News" locale: "en" url: "https://longbridge.com/en/news/289152652.md" description: "A latest NBER paper from Wharton School reveals a stark arithmetic reality: tech giants like Amazon, Microsoft, and Meta Platforms are projected to reach $755 billion in AI capital expenditure in 2026, potentially exceeding $1.1 trillion in 2027. The only \"solution\" to this high-stakes gamble is a 2.7-fold productivity leap within five years; otherwise, they face receivership risks. Meanwhile, DeepSeek is challenging GPT at one-tenth the cost, subjecting the high-premium logic of frontier labs to a double squeeze" datetime: "2026-06-09T07:34:38.000Z" locales: - [zh-CN](https://longbridge.com/zh-CN/news/289152652.md) - [en](https://longbridge.com/en/news/289152652.md) - [zh-HK](https://longbridge.com/zh-HK/news/289152652.md) --- # Wharton Paper Warns: AI Must Boost Productivity by 2.7x to Avert Tech Receivership A new study quantifies the high-stakes AI gamble of tech giants into a matter of life-and-death arithmetic: either achieve an unprecedented productivity leap or face the risk of receivership. According to the latest working paper released by the National Bureau of Economic Research (NBER), Jessica Wachter, a financial economist at Wharton School, and Jonathan Wachter from the hedge fund Point72, jointly modeled and calculated that the combined capital expenditure of Amazon, Alphabet, Microsoft, Meta Platforms, and Oracle will reach $381 billion in 2025, and is expected to climb to approximately $755 billion in 2026—more than triple the 2024 level. The authors estimate that this figure will further rise to about $1.1 trillion in 2027. The core conclusion of the paper points directly to the internal logic of this gamble: **the scale of such expenditure is only justified if the AI industry achieves a productivity leap of approximately 2.7 times; otherwise, the relevant companies will "face receivership risks."** Meanwhile, AI investment already accounts for an estimated 14% of all private fixed investment in the United States and 2.4% of GDP, surpassing the peak level of about 1.5% during the telecommunications investment boom of the late 1990s. Its magnitude is now sufficient to influence the overall economic trend. ## Unprecedented Scale of the Bet The paper employs a "rare productivity boom" model to reverse-engineer the spending behavior of tech giants, attempting to clarify the expectations implied behind this investment frenzy. The results show that achieving a 2.7-fold productivity increase within about five years would surpass any record in comparable periods in economic history. The paper cites two historical benchmarks for illustration: **it took the US railroad era about 60 years to nearly triple per capita GDP; and throughout the entire IT boom cycle from 1995 to 2005, the productivity increase was only 1.5 times. By this measure, the speed and magnitude of the current bet by tech giants are unique cases.** If the bet pays off, the model predicts that cumulative US GDP growth by 2030 will receive an additional boost of 5 to 58 percentage points. The wide range of this interval itself reflects the high uncertainty of the forecast. ## AI Investment Has Become an Economic Pillar Data from the paper reveals that the penetration of AI investment into the US macroeconomy has far exceeded general market perception. **In the second half of 2025, AI contributed about one-fifth of real GDP growth; if AI-related expenditures were excluded, overall corporate equipment investment would show negative growth.** In terms of proportion, the share of AI investment in US private fixed investment has jumped from 3.3% in 2022 to an estimated 14%, and its 2.4% share of GDP has surpassed the peak level of telecommunications investment during the previous tech bubble. This means that any significant contraction in AI investment will have a quantifiable downward impact on the overall economy. The authors of the paper clearly defined the boundaries of their conclusions, and investors should note two important limitations when interpreting them. First, the forecast of approximately $1.1 trillion in capital expenditure for 2027 is an estimate made by the authors based on a bottom-up approach. Currently, no company has issued capital expenditure guidance for 2027, so this figure does not come from official corporate disclosures. Second, "receivership risk" is an argument based on "revealed preference"—meaning that given the current scale of expenditure, a significant productivity increase is a necessary condition to make these expenditures rational—rather than a prediction that the aforementioned companies will actually go into receivership. In other words, the paper describes the internal logic of this gamble, not a judgment on the outcome. ## Cheap Alternative Models Intensify Pressure on Frontier Labs Against this macroeconomic backdrop, frontier AI labs also face direct challenges from low-cost competitors. **An independent test compared DeepSeek V4 Pro with GPT-5.5 Pro on precise tasks, showing that DeepSeek won 38 to 33 in four text tests, with the tester using Grok as the judging model.** However, the methodology of this test was widely questioned on Hacker News: critics pointed out that the experiment covered only four poorly designed tasks, lacked a reproducible process, and the judging model used had been discontinued. Nevertheless, the cost gap consistently reported by users remains a more substantive signal—commenters noted that in another vulnerability scanning test, the operating cost of DeepSeek was about one-tenth that of GPT Pro. This controversy reflects deeper structural pressures in the market: as corporate customers accelerate their search for ways to compress token costs and increasingly tend to adopt open-source alternatives like DeepSeek, the pricing strategies and investor narratives of frontier labs are being tested. The productivity threshold revealed by the Wharton paper is the ultimate quantitative expression of this pressure—if the assertion that cheap models are "good enough" withstands scrutiny, the logic underlying the high premiums maintained by frontier labs will face fundamental challenges. ### Related Stocks - [MSFT.US](https://longbridge.com/en/quote/MSFT.US.md) - [META.US](https://longbridge.com/en/quote/META.US.md) - [AMZN.US](https://longbridge.com/en/quote/AMZN.US.md) - [FBL.US](https://longbridge.com/en/quote/FBL.US.md) - [AMZD.US](https://longbridge.com/en/quote/AMZD.US.md) - [AMZU.US](https://longbridge.com/en/quote/AMZU.US.md) - [MSFX.US](https://longbridge.com/en/quote/MSFX.US.md) - [METW.US](https://longbridge.com/en/quote/METW.US.md) - [METU.US](https://longbridge.com/en/quote/METU.US.md) - [MSFU.US](https://longbridge.com/en/quote/MSFU.US.md) - [MSFL.US](https://longbridge.com/en/quote/MSFL.US.md) - [AMZP.US](https://longbridge.com/en/quote/AMZP.US.md) - [MSFO.US](https://longbridge.com/en/quote/MSFO.US.md) - [DPSK.NA](https://longbridge.com/en/quote/DPSK.NA.md) - [METD.US](https://longbridge.com/en/quote/METD.US.md) - [MSFD.US](https://longbridge.com/en/quote/MSFD.US.md) - [AMZZ.US](https://longbridge.com/en/quote/AMZZ.US.md) - [AMZW.US](https://longbridge.com/en/quote/AMZW.US.md) - [MSFW.US](https://longbridge.com/en/quote/MSFW.US.md) - [MSFY.US](https://longbridge.com/en/quote/MSFY.US.md) - [FBY.US](https://longbridge.com/en/quote/FBY.US.md) - [FBYY.US](https://longbridge.com/en/quote/FBYY.US.md) - [AMZO.US](https://longbridge.com/en/quote/AMZO.US.md) - [AMZY.US](https://longbridge.com/en/quote/AMZY.US.md) - [AZYY.US](https://longbridge.com/en/quote/AZYY.US.md) - [GOOGL.US](https://longbridge.com/en/quote/GOOGL.US.md) - [GOOG.US](https://longbridge.com/en/quote/GOOG.US.md) - [ORCL.US](https://longbridge.com/en/quote/ORCL.US.md) - [ORCL-D.US](https://longbridge.com/en/quote/ORCL-D.US.md) ## Related News & Research - [Meta Investor Dumps Shares After AI Fundraising Report, Says Shareholders Don't Want 'Drunken Sailors' Spending: 'People Are Not Going To Invest...'](https://longbridge.com/en/news/288922790.md) - [Meta funds skilled trades jobs program for AI data center buildout](https://longbridge.com/en/news/289104022.md) - [ANALYSIS-High-profile Meta AI chatbot breach spotlights security risks of automation](https://longbridge.com/en/news/288554153.md) - [Meta made its own AI-generated clickbait news feed](https://longbridge.com/en/news/288934190.md) - [3 Artificial Intelligence (AI) Stocks to Buy and Hold for the Next Decade](https://longbridge.com/en/news/288892529.md)