US Stocks Rally Strongly, Philadelphia Semiconductor Index Gains Over 2%: Chip Stocks and Cloud Providers Resume Rise; Analysts Say Kimi K3 Selloff Irrational

TradingKey - On July 20, Eastern Time, the three major US stock indexes diverged once again, with tech stocks resuming their upward momentum. The Dow Jones Industrial Average fell, while the S&P 500 and the Nasdaq Composite Index rose. The Philadelphia Semiconductor Index gained over 2%, with 26 of its 30 components advancing.

As of press time, the Dow Jones Industrial Average fell 0.18% to 52,051.10 points; the Nasdaq Composite Index rose 0.82% to 25,728.81 points; and the S&P 500 Index rose 0.41% to 7,488.09 points.

Source: FutuBull

From a sector perspective, chip stocks and memory stocks rebounded after bottoming out; among them, SanDisk ( SNDK ), Marvell Technology ( MRVL) rose about 6%, Micron ( MU) rose 5%, Intel ( INTC ), AMD ( AMD) and Broadcom ( AVGO) all rose over 3%.

Notably in today's trading, hyperscale data center stocks and AI hardware stocks gained together, a shift from last month's capital flows. Google ( GOOGL) rose 2.38%, Microsoft ( MSFT) rose 2.21%, Amazon ( AMZN) rose 2.11%, Meta Platforms ( META) rose 1.00%.

Just last week, the market worried that the growth logic for AI computing demand would be undermined due to the Kimi K3 model's performance entering the global tier-one and its superior computing efficiency, triggering a collective pullback in the AI infrastructure chain.

Morningstar analysts countered the market panic, arguing that the sell-off was an emotionally driven irrational reaction. They emphasized that improved efficiency in AI algorithms is fundamentally a long-term positive for the industry and does not shake the core investment logic of cloud vendors.

The firm explained this through the following three logical points.

First, technological iteration itself is the core fundamental support of the industry. K3's performance leap through architectural and algorithmic innovations precisely demonstrates that AI technology continues to make breakthroughs. For the entire AI industry, the true systemic risk is the stagnation of technological iteration, rather than the cost reductions brought about by improved efficiency.

Second, regulatory barriers make it difficult for Western companies to shift to Chinese models on a large scale. Driven by national security and technology control considerations, it is almost impossible for the US government to allow US and Western companies to tilt their core computing spending toward Chinese open-source models. Thus, the scenario of a massive shift in demand feared by the market lacks a solid foundation.

Third, improved efficiency will not shrink the total demand for computing power, but will instead expand the market pie. Lower inference costs will lower the barrier to entry for AI usage, spawn more application scenarios and call volumes, and drive the continuous expansion of overall computing demand. Cloud vendors, as infrastructure providers, will remain the core beneficiaries.

From a product perspective, although K3 has broken the performance boundaries between open-source and closed-source models with its 2.8 trillion parameters, million-level context window, open-source weights, and outstanding coding/agent capabilities, its overall capabilities have not yet surpassed the top closed-source models in the Claude and GPT series, and it has not yet established a disruptive generational advantage.

Reportedly, in the Arena AI Front-End Coding Arena, Kimi K3 topped the global rankings with 1,679 points, exceeding Claude Fable 5's 1,631 points and GPT-5.6 Sol's 1,618 points. Among the seven sub-sectors of brand marketing, reference design, data analysis, consumer products, simulation, content creation, and gaming, K3 secured first place in six, trailing slightly behind Fable 5 only in the gaming category.

Goldman Sachs stated that Chinese open-source AI models "are reaching a critical tipping point of intelligence for global diffusion." K3 raised API pricing to the highest level among Chinese models, but it remains only about one-third of Claude Fable 5's.

Morgan Stanley stated that it positions K3 as a signal that Chinese LLMs are "catching up comprehensively with US leaders in model size, performance, and pricing." The firm added that it does not view K3 as an overnight miracle, but rather the result of cumulative progress in China's AI industry. The report also expects more Chinese models with 2 to 5 trillion parameters to be released in the second half of the year.