by Krystal Daibes Higgins, CFA
Vice President
Equity Research
The AI Race Continues
This week, the equity market had a “shoot first, ask questions later” response to news surrounding DeepSeek, a Chinese artificial intelligence startup company that claimed to achieve ChatGPT-level performance at a fraction of the cost. This news sent a shockwave through the technology sector, sparking a frenzy of speculation and questions about AI innovation. Specifically, the revelation raised concerns about the capital requirements for developing high-quality AI models and triggered a wave of panic selling. Companies at the forefront of AI advancement took huge hits to their stock prices, mainly semiconductor companies, utilities and other companies benefiting from data center buildouts.
Industry experts are scrutinizing DeepSeek’s claims, hoping better to understand the full picture behind their reported success. For example, the company reported that it only cost $6 million to train its most recent model, excluding “costs associated with prior research” and other advancements. Most experts consider this claim highly unlikely. The biggest debate lies in how many and what kind of graphic processing units (GPUs) were used. GPUs are semiconductors that enable powerful computing processing to train and run large language models such as AI. As the week progressed, we saw some reversal of panic selling, but uncertainties persist regarding the implications for companies pioneering the infrastructure buildout of AI.
At Ferguson Wellman, we continue to seek more clarity on the reports debunking DeepSeek’s claims; however, this debate is not at the crux of our investment theses. Our belief is that, regardless of the cost, power and amount of energy it took to train DeepSeek’s model, the heavy capital expenditures will continue as the “AI race” intensifies and remains in the beginning stages.
On Wednesday, the companies investing the most in AI reported their earnings and provided a business update. These reports corroborated our belief that significant investments will continue, which should benefit the companies’ share prices that were hurt the most this week.
Microsoft reiterated that it plans to maintain its capital expenditure of approximately $80 billion in its fiscal year on AI investments and that a greater percentage of that amount will be allocated to GPUs and computer servers and less to long-term assets such as buildings and land.
Similarly, Meta, the company that owns Facebook and Instagram, also reaffirmed that it will be spending “hundreds of billions of dollars” on AI infrastructure over the long term. Its CEO, Mark Zuckerberg, refuted DeepSeek as a negative for capital expenditures as he believes “investing very heavily in [capital expenditures] and infrastructure is going to be a strategic advantage over time.”
A Volatile Week
The equity markets proved volatile with all the noise surrounding DeepSeek, causing the S&P 500 to end slightly negatively. The barrage of news, plus the reporting of some of the largest companies in the world, is moving the markets underneath the surface. This week, all but two of the Magnificent 7 reported earnings, including Amazon, Apple, Meta, Microsoft and Tesla. These five provided updates largely in line with expectations, although their stock prices had different reactions. Fundamentals appear to remain strong and on track to deliver the expected earnings growth in 2025. As for the remaining two, Alphabet is set to report next week, and Nvidia, which was in the eye of this week’s AI storm, will be the last of the group to report at the end of February. Until Nvidia reports, and very likely even after, we expect to see continued volatility as the market digests all the reports on new AI developments.
Takeaways for the Week
The equity market reacted negatively to DeepSeek's claims of achieving ChatGPT-level performance at a fraction of the cost, causing significant volatility in the technology sector, particularly affecting semiconductor companies and those involved in data center buildouts
Despite the uncertainties surrounding DeepSeek's claims, major companies like Microsoft and Meta reaffirmed their commitment to substantial investments in AI infrastructure