From 1929 to 2026: The K-shaped Fallacy of AI

```

Global AI computing power stocks saw a sharp decline mid-year, reigniting market concerns over the "K-shaped divergence" between technology and consumption.

In a research report from June 13, Western Securities’ strategy team argued that such concern both misjudges the real reasons for the current downtrend in computing power and misinterprets the historical patterns of technological waves and wealth distribution.

According to the report, the recent correction in computing power stocks—not only a result of high US bond yields suppressing the growth style—may fundamentally stem from overly high barriers to use caused by AI large models "stacking parameters," which has temporarily discouraged ordinary users and raised fears of insufficient computing power demand. Meanwhile, the narrative that "AI worsens wealth disparity" is also spreading in the market. The report firmly refutes this: the establishment of global welfare societies after WWII fundamentally changed the relationship between technological innovation and wealth distribution. The semiconductor wave in the US and Japan in the 1970s already provides a historical example—technology and consumption can resonate, not diverge.

The report argues that improved macro conditions are the foundation for the long-term advance of technology. For sector allocation, it recommends maintaining a "bull market barbell" strategy, balancing AI computing power volatility with opportunities in consumption recovery.

Computing Power Stocks Drop Sharply: Supply-side Worries Surface

Global AI computing power sectors experienced a significant mid-year pullback. Western Securities believes the trigger is not limited to high US bond yields suppressing growth styles, but more deeply lies in market concerns over a temporary lack of computing power demand.

The report notes that in the early stages of AI large models, various users rushed in and computing power demand surged rapidly. However, as large models evolved into "parameter stacking," the technical threshold kept rising, deterring some ordinary users and causing user growth to hit a temporary bottleneck. Combined with Meta's previous disclosures about possibly selling excess computing power in the future, worries about a supply-demand mismatch in computing power further fermented in the market.

The report sees the solution as AI shifting from "parameter stacking" to "lowering barriers." For instance, the rapidly growing WorkBuddy focuses on lowering the usage threshold as its main selling point, ranking first among desktop AI office agents in March 2026 with 8.85 million monthly visits and a month-over-month growth rate of 831%. The report argues that when more ordinary, non-technical users can easily use AI, global computing power demand may usher in another round of accelerated expansion.

The K-shaped Fallacy: History Has Twice Debunked This Narrative

Some investors pit "carbon-based" consumption against "silicon-based" technology, believing the AI wave will inevitably intensify "K-shaped" economic divergence. Western Securities qualifies this as "a misunderstanding of historical patterns."

Reviewing history, the report notes that before WWII, there was indeed a marked K-shaped divergence between technology development and consumption trends: in early 20th-century America, with heavy industrialization and electrification, hydroelectric generators, steam commercial generators, and Ford assembly lines appeared one after another, but the technological dividends failed to effectively reach household consumption. At the time, the lack of distribution mechanisms meant technological innovation increased total wealth supply, but could not boost demand synchronously—eventually, this laid the seeds for the 1929 Great Depression.

However, circumstances fundamentally changed after WWII. Welfare societies gradually formed worldwide, creating a new setup of "technological innovation grows the pie, welfare society divides the pie." In the US semiconductor wave of the 1970s, military tech accumulated during WWII entered mass consumption, chips and processors saw breakthroughs, and with continuing welfare improvements, residents shared in technological progress, resulting in consumption and technology rising together. Japan also implemented the "VLSI (Very Large Scale Integration) Plan" at the same time, and similarly saw technology and consumption prosper hand in hand.

Welfare Societies: The Institutional Force Correcting Wealth Distribution

The postwar US and Japanese experience shows that government welfare spending played a crucial re-distributive role. The report’s data shows that social security and welfare in both countries after WWII grew much faster than wages, allowing purchasing power to keep pace with productivity, thus fostering a virtuous resonance between technological development and consumption prosperity.

The current AI wave is sparking similar policy discussions. As noted in the report, the South Korean government, US senators, and OpenAI CEO Sam Altman have all publicly discussed the “citizens’ dividend” concept, all pointing to using AI-generated added wealth to benefit all citizens.

The report extends the same logic to China. It notes that by around 2018, China had formally entered a mature stage of industrialization; since 2020, correcting factor prices and "expanding domestic demand" have become policy themes, aligning closely with postwar US and Japanese historical paths. Thus, the report concludes that, in the long run, incremental wealth from AI in China is expected to benefit residents more broadly—AI will not intensify the "K-shaped divergence," but rather may speed up the "common prosperity" process.

Macro Conditions Improvement is the Fundamental for Long-Term Technology Advance

Western Securities further elaborates that higher factor prices (inflation environments) are conducive to cutting-edge technological innovation: on one hand, high costs force companies to increase R&D spending; on the other, high household incomes help firms turn R&D outcomes into real sales. The report concludes: "Macro-economic improvement, corporate profit recovery, and structurally optimized finance are needed for technological innovation to go further and more steadily."

Sector Allocation: Barbell Strategy to Cope with High Volatility

On the investment strategy side, the report advises keeping the "bull market barbell" framework during the period when AI computing power demand is under short-term pressure, and takes a fairly positive view on a Q3 consumption rebound.

Specifically, the report suggests two allocation approaches: First, growth investors must accept the high-volatility nature of AI computing power hardware (communication equipment, semiconductors, memory, etc.) as one end of the barbell; Second, the consumption sector is the other end of the barbell and approaches are scenario-dependent—if household balance sheets significantly repair in the second half of the year, focus on consumption reversal opportunities, especially in real estate and baijiu (liquor); if household balance sheet repair falls short, then while capturing consumption rebound, also pay attention to PPI export chain-related sectors, such as new energy, chemicals, pharmaceuticals, and home appliances.

Risk Disclosure and DisclaimerThe market has risks; investment should be cautious. This article does not constitute individual investment advice, nor does it consider any individual user's specific investment goals, financial status, or needs. Users should consider whether any opinions, views, or conclusions in this article suit their own situation. Investment decisions based on this article are at your own risk. ```