--- title: "\"Is 'Kimi Moment' actually beneficial for chips? Wall Street: The 'Jevons Paradox' will be proven again!" type: "News" locale: "en" url: "https://longbridge.com/en/news/293276058.md" description: "The release of the Chinese AI model Kimi K3 has sparked a debate on Wall Street regarding the rationality of computing power expenditures. Despite the decline in the Philadelphia Semiconductor Index, analysts believe that low-cost Chinese AI may benefit American chip giants through the \"Jevons Paradox,\" as increased demand ultimately benefits chip companies" datetime: "2026-07-21T02:20:47.000Z" locales: - [zh-CN](https://longbridge.com/zh-CN/news/293276058.md) - [en](https://longbridge.com/en/news/293276058.md) - [zh-HK](https://longbridge.com/zh-HK/news/293276058.md) --- # "Is 'Kimi Moment' actually beneficial for chips? Wall Street: The 'Jevons Paradox' will be proven again! **Financial Associated Press, July 21 (Editor: Xiao Xiang)** Last week, a highly competitive AI model from China across the ocean reignited the debate on Wall Street about whether the massive spending on storage components and computing power in the U.S. AI industry is justified—the Philadelphia Semiconductor Index also fell into a technical bear market last Friday. **However, after a weekend of heated discussions, some Wall Street analysts and U.S. tech industry insiders believe that the emergence of low-cost Chinese AI may not necessarily be a bad thing for American chip giants, even though the stock prices of these chip giants had previously soared due to strong external demand for AI hardware.** Their reasoning is that—even if AI costs decrease, these chip companies may ultimately benefit from it. Last Thursday, Chinese AI developer Moonlight released its latest Kimi K3 model, an open-weight model that ranks comparably to leading U.S. models in certain benchmark tests. In the Arena comprehensive text ranking, Kimi K3 currently ranks first in the front-end code leaderboard with a score of 1679, ahead of Anthropic's Claude Fable 5 (1631 points) and OpenAI's GPT-5.6 Sol (1618 points)—both of which are closed-source models. This ranking is based on approximately 484,000 pairwise blind test votes conducted on 98 models. Given that K3's predecessor ranked 18th, Moonlight has achieved a significant leap in performance in a very short time. A Chinese laboratory, with limited funding resources and facing U.S. export controls on Nvidia's latest and most powerful chips, has managed to develop a model with near-frontier capabilities, which has sent shockwaves through Wall Street and Silicon Valley. The release of Kimi K3 intensified the sell-off in the U.S. chip sector last Friday. However, some industry insiders stated that **the soaring demand for Kimi K3 indicates that Chinese AI models may be on par with their Western counterparts in terms of computing power hunger.** Moonlight stated over the weekend that the explosive demand for Kimi K3 has pushed its capacity to the limit, leading the company to suspend new user subscriptions to prioritize the allocation of computing power for existing members. **Will the Jevons Paradox be proven again?** **Anni Sen, Managing Partner of BluBird Capital, clearly stated that this dynamic is "absolutely a tailwind for memory chip trading."** Data from the open-source community Hugging Face shows that among Kimi K3's total parameters of up to 2.8 trillion, only about 50 billion are active parameters. Parameters are variables used by AI models during the training phase to learn and recognize feature patterns, equivalent to the model's "memory units." Anni Sen explained that while this design greatly enhances the operational efficiency of the model, having 28 trillion parameters reside in main memory in real-time means that ordinary enterprises can hardly run the model independently on their own servers. Kimi K3 currently charges $15 per million output tokens—by comparison, OpenAI's GPT-5.6 Sol charges $30, and Anthropic's Claude Fable 5 charges $50. **Wall Street professionals believe that, as an embodiment of the "Jevons Paradox"—which states that improving the efficiency of a resource may increase its total demand, the lower token cost of Kimi K3 may lead to an increase in AI workloads.** **Sen believes that "as developers leverage its lower cost structure, Kimi will create more application scenarios," which will further increase the demand for inference. Sen stated that enterprises will increasingly migrate "mid-tier" AI tasks to cheaper models while reserving extremely expensive cutting-edge models for more complex multi-step intelligent agent processes.** AI researcher and founder of the Interconnects AI blog Nathan Lambert wrote in an article on Monday: "Open-weight models accelerate the proliferation of AI across the economy by lowering the intelligence threshold at specific performance levels." He added that open models provide additional customization opportunities, making them more useful for specific enterprises. **Chip demand will further increase** **Wedbush analyst Matt Bryson stated that as AI models become larger, the demand for memory to support more parameters will also grow.** Bryson noted in a report on Monday that this means the memory capacity of individual AI chips must be significantly increased, or it will require expanding the scale of chip clusters to share the burden of carrying large weight data. Therefore, Bryson stated, **if Chinese AI models continue to gain favor, considering the increased demand for high-performance memory, this could be "a good thing" for memory suppliers. Additionally, the increase in chip cluster size due to more model parameters may also provide a boost to network equipment providers.** Micron Technology, SK Hynix, and Samsung Electronics are currently the three major suppliers of high-bandwidth memory (HBM) globally, and the shortage of memory components has given these companies significant pricing power. D.A. Davidson Managing Director Gil Luria stated that as the cost of running AI models decreases, the demand for computing power may increase, benefiting the chip manufacturing industry—especially in the area of memory chips, which are currently experiencing supply tightness. PurePlay ETFs CEO Joseph DeYonker also expressed a similar view, stating that **as AI models become more efficient and accessible, the surge in user adoption "will immediately test the limits of physical infrastructure"—as is the case with Kimi K3 In his view, this confirms that the long-term growth trajectory of chip manufacturers remains intact.** DeYonker stated, "Running advanced frontier models with large context windows is putting unprecedented pressure on the memory and chip manufacturing supply chain." Therefore, he expects companies focused on HBM and advanced chip packaging to maintain strong pricing power. **At the same time, DeYonker also anticipates that designers of GPUs and custom AI chips—such as NVIDIA and Broadcom—will continue to see strong order backlogs** as enterprises and hyperscale computing providers scramble to acquire the necessary hardware to avoid specific operational bottlenecks. **In his view, the restrictions on Kimi K3 subscriptions are "a structural confirmation of the semiconductor industry, rather than a cyclical warning signal,"** and investors should view the ongoing memory and computing bottlenecks as signs of the chip sector's long-term sustainable profit growth potential ### Related Stocks - [OpenAI.NA](https://longbridge.com/en/quote/OpenAI.NA.md) - [NVDA.US](https://longbridge.com/en/quote/NVDA.US.md) - [MU.US](https://longbridge.com/en/quote/MU.US.md) - [SSNGY.US](https://longbridge.com/en/quote/SSNGY.US.md) - [AVGO.US](https://longbridge.com/en/quote/AVGO.US.md) - [NVD.DE](https://longbridge.com/en/quote/NVD.DE.md) - [SMSN.UK](https://longbridge.com/en/quote/SMSN.UK.md) ## Related News & Research - [How the 'Nostradamus of AI' failed to spot a market crash](https://longbridge.com/en/news/294580418.md) - [Meet the Gen Z AI whiz at the center of a hedge fund meltdown](https://longbridge.com/en/news/294425537.md) - [These 2 ETFs Let You 2X Chip Stocks No Matter Which Direction They Head Next](https://longbridge.com/en/news/294379303.md) - [EXCLUSIVE: Elon Musk's Space Data Center Plans Are Far From 'Dumb,' Says Gates-Backed Chip Founder](https://longbridge.com/en/news/293959858.md) - [Outside The Semi-mageddon, Chipocalypse & Magnificent Meltdown, The Rest Of The Market Looks Pretty Wonderful...For Now](https://longbridge.com/en/news/294184795.md)