The AI landscape is undergoing a dramatic transformation as DeepSeek’s efficient and cost-effective model challenges Big Tech’s dominance, forcing investors to reassess valuations and pricing power across AI-related stocks. This shift from proprietary, resource-intensive models to more accessible “open weight” alternatives is creating both volatility and opportunity in the market.
DeepSeek’s impact has been immediate and profound. The Chinese AI model demonstrated that optimization is primarily a software challenge rather than a hardware problem, proving that massive supercomputers aren’t the only path to AI efficiency. Unlike fully open-source models, DeepSeek is “open weight,” sharing its memory and knowledge processes while keeping full code proprietary. This approach lowers barriers for startups and companies to create specialized business models, potentially driving explosive growth across the sector.
The market reaction has been swift and sometimes indiscriminate. ARM Holdings, despite reporting exceptional earnings, saw its stock drop over 7% as investors grappled with uncertainty. However, ARM CEO Rene Haas embraced DeepSeek during the company’s February 5 earnings call, calling it “creative” and predicting it would increase computing demand, particularly for smaller devices that don’t require high-powered Nvidia chips.
JPMorgan Private Bank’s North America equity research team has identified key winners in this new competitive landscape, focusing on companies positioned to benefit from AI efficiency rather than just raw computing power. Abby Yoder, US equity strategist at JPMorgan, emphasizes that while CapEx spending continues to increase—with Amazon projected at $97 billion, Meta at $65 billion, and Google at $62 billion for 2025—the critical question is where that money flows.
JPMorgan’s top picks span four categories:
Software companies like Snowflake (SNOW) and Salesforce (CRM) are positioned to benefit from increased demand as AI applications proliferate. Semiconductor firms including Broadcom (AVGO), Marvell Technologies (MRVL), and Micron (MU) are expected to gain from demand for application-specific integrated circuits (ASICs) and memory chips. Internet hyperscalers—Alphabet, Amazon, and Meta—remain crucial as both AI developers and infrastructure spenders. Meta’s open-weight Llama model positions it particularly well, while Amazon’s AWS has already integrated DeepSeek into its platforms. Hardware companies like Dell (DELL), Cisco (CSCO), and NetApp (NTAP) stand to benefit from the inference process as AI model usage accelerates.
Key Quotes
It’s clearly evident that this stock is supposed to be soaring. The company’s IP is used in virtually every major smartphone in the world. Its software is also used by Apple computers. It has a tie-up with Nvidia, and its backup is SoftBank, which is part of the Stargate project pumping $500 billion into AI cloud infrastructure.
Sandeep Rao, quantitative analyst at Leverage Shares ETPs, commented on ARM’s paradoxical stock decline despite stellar earnings, highlighting how market uncertainty about AI efficiency is affecting even fundamentally strong companies with extensive industry partnerships.
DeepSeek changed that thesis, suggesting that optimization isn’t a hardware problem but one that can be addressed with code. It also proved that the open-source community was far more efficient at driving results, validating the argument that AI and machine learning should not be done behind closed doors.
Rao’s analysis captures the fundamental shift DeepSeek represents—challenging the prevailing assumption that AI advancement requires ever-more-powerful hardware and instead demonstrating that software optimization and collaborative development can achieve superior results.
So, are they going to be going to the incumbent winners that we’ve seen over the past couple of years who have had this outsized pricing power leading to those really high margins? Or are they going to semiconductor companies making specialized chips, hardware providers, or software companies?
Abby Yoder, US equity strategist at JPMorgan Private Bank, frames the critical investment question facing the market: whether capital will continue flowing to established AI infrastructure providers or shift to more specialized, efficient alternatives.
At 4Q earnings, META reiterated the benefits of open source LLMs, noted that costs can be driven down, & — in light of DeepSeek — emphasized the importance of having an American standard for open source that extends globally.
JPMorgan’s research note on Meta highlights how the company’s open-weight Llama model positions it advantageously in the efficiency era while also touching on the geopolitical dimension of AI development standards.
Our Take
The DeepSeek disruption represents more than a temporary market correction—it’s a fundamental challenge to the AI infrastructure thesis that has driven trillions in market capitalization. What’s particularly striking is how quickly sentiment shifted from “bigger is better” to questioning whether expensive, proprietary models justify their premium pricing. JPMorgan’s strategic pivot toward efficiency beneficiaries is telling: they’re not abandoning AI infrastructure plays but rather refining their thesis to focus on adaptable companies that can thrive regardless of whether the future is dominated by massive proprietary models or distributed, efficient alternatives. The inclusion of both hyperscalers and specialized chip makers suggests a hedge—acknowledging that the ultimate winners may come from multiple categories. Most intriguing is the emphasis on software companies like Snowflake and Salesforce, which could see their total addressable markets expand dramatically if AI becomes cheaper and more accessible. This shift from hardware-centric to software-centric value creation mirrors previous technology transitions and suggests we’re entering AI’s pragmatic phase, where real-world applications and business value matter more than raw computational power.
Why This Matters
This development marks a pivotal inflection point in AI’s evolution from a capital-intensive race dominated by a few tech giants to a more democratized, efficiency-focused ecosystem. DeepSeek’s emergence validates the argument that collaborative, open development can outpace closed proprietary systems, potentially reshaping the entire AI value chain.
For investors, this shift demands a fundamental reassessment of where value will be created. The traditional assumption that expensive, proprietary models with massive computing requirements would dominate is being challenged. Companies that can leverage efficiency to expand their total addressable markets—rather than those simply hoarding computing power—are positioned to win.
The broader implications extend beyond stock prices. More accessible AI models could accelerate adoption across industries, enabling smaller companies and startups to compete with established players. This democratization could drive innovation in healthcare, education, manufacturing, and countless other sectors previously priced out of advanced AI capabilities. However, it also raises questions about American technological leadership, as JPMorgan notes the importance of establishing “an American standard for open source that extends globally.” The competition between efficiency and raw power will define the next chapter of AI development.
Recommended Reading
For those interested in learning more about artificial intelligence, machine learning, and effective AI communication, here are some excellent resources:
Recommended Reading
Related Stories
- The AI Hype Cycle: Reality Check and Future Expectations
- Wall Street Asks Big Tech: Will AI Ever Make Money?
- Legendary Investor Jeremy Grantham Predicts Stock Market Crash and AI Bubble Burst by 2024
- Encharge AI Secures $21.7M Series A Funding to Revolutionize AI Chip Efficiency
- Jensen Huang: TSMC Helped Fix Design Flaw with Nvidia’s Blackwell AI Chip