2024 U.S. stock market review: The biggest "dark horse" surged 741% annually, Nvidia was counterattacked in Q4, Microsoft ranked last among the "Magnificent Seven", institutions warn of AI bubble

Looking back at 2024, amid the AI boom, the three major U.S. stock indices continued their upward trend from the previous year and repeatedly hit record highs. The S&P 500 index closed at a record high nearly 60 times during the year, the Dow Jones Industrial Average surpassed 45,000 points for the first time, and the Nasdaq Composite Index broke through 20,000 points for the first time in history.
As of the close on December 27, the S&P 500 index stood at 5,970.8 points, up 25.18% year-to-date; the Dow Jones Industrial Average was at 42,992.21 points, up 14.07%; and the Nasdaq Composite Index was at 19,724.66 points, up 31.38%. Although the "Magnificent Seven" tech giants remained strong, their performance diverged significantly. Driven by strong demand for AI chips, Nvidia led the "Magnificent Seven" with a 176% annual gain. However, Broadcom is stealing some of its thunder. Microsoft, which owns OpenAI, saw only a 15% annual gain. The "Magnificent Seven" did not have the best performance this year. Two companies stood out due to their AI applications. Among them, AppLovin emerged as the biggest "dark horse," with a 741% annual gain, making it the top performer among the components of the Dow Jones, S&P 500, and Nasdaq 100 indices. Palantir's stock price rose 360% this year, making it the best performer in the S&P 500 index.

Year-to-date, the stock price of gaming and digital advertising company AppLovin has surged over 741%, making it the top performer among the components of the Dow Jones, S&P 500, and Nasdaq 100 indices. The driving force behind AppLovin's stock surge is its AI advertising engine, Axon. In the third quarter of this year, AppLovin's software platform revenue, including the Axon engine, grew 66% year-over-year to $835 million. The company's net income skyrocketed 300%, with margins improving from 12.6% a year ago to 36.3%. Tim Nollen, a senior equity analyst at Macquarie, believes that AppLovin's e-commerce initiatives centered around the Axon engine could unlock a potential advertising revenue opportunity of approximately $120 billion. However, some analysts remain cautious about the sustainability of AppLovin's advertising growth. Similarly, big data analytics company Palantir has gained significant attention due to its AI applications. In April 2023, Palantir launched a software suite called the AI Platform (AIP). AIP has helped Palantir secure partnerships with tech giants like Microsoft, Oracle, Meta, and Amazon. Palantir's stock price has risen 360% year-to-date, making it the "champion" among S&P 500 components. Dan Ives, an analyst at WedBush, believes Palantir could become "the next Oracle." He rates Palantir as "outperform" with a target price of $75. As of press time, Palantir's stock price was $79.08. However, Wall Street is generally pessimistic about Palantir's prospects in 2025. Among the 22 analysts surveyed by FactSet, only three are bullish, while seven are bearish. The average target price for the company is $44.87, 43% below the current price.

In terms of performance, Nvidia remains the biggest beneficiary of the AI boom. Although its gains are not as high as AppLovin's or Palantir's, Nvidia's stock price has risen 176% year-to-date after a 239% surge last year, bringing its market capitalization to $3.35 trillion.
In the AI chip market, Broadcom is stealing some of the spotlight with its ASICs (Application-Specific Integrated Circuits), which are custom-designed chips tailored to specific user or system needs. The insatiable demand for processing power from major cloud service providers like Google, Microsoft, and Amazon has created a huge market opportunity for Broadcom. Year-to-date, Broadcom's stock price has risen 116%. In the fourth quarter alone, its stock surged 43%, overshadowing Nvidia's 12.57% gain.

Tianyu Jia, an assistant professor at Peking University's School of Integrated Circuits, stated that ASICs or domain-specific chip designs are considered an important trend by Turing Award winners David Patterson and John Hennessy, as they offer significant advantages in reducing chip power consumption and costs. In high-performance AI computing, power consumption is a major limiting factor for chip scalability. Compared to GPU chips, custom ASIC designs can deliver substantial energy efficiency and cost advantages. Morgan Stanley analysts predict that by 2027, ASICs will account for 13% of all AI accelerator sales, up from 11% this year. By 2030, ASICs could reach 15%. However, Jia noted: "Building a robust ASIC chip development ecosystem requires enormous effort, so replacing GPU chips still has a long way to go."

CITIC Securities stated that traditional general-purpose GPU vendors like Nvidia will face significant challenges as many customers are developing their own solutions or turning to Broadcom to reduce the total cost of deep neural network inference. Bank of America analysts predict that by 2030, Nvidia will still hold 75% of the market share, slightly down from 80% in 2024. Morgan Stanley also believes Nvidia will maintain its dominant position and has named it the "top pick" for next year, with a target price of $166 per share.
This year, the "Magnificent Seven" continued their upward trend, with Nvidia leading the pack and Tesla following closely behind. Surprisingly, Microsoft, which invested in OpenAI, lagged with only a 15% gain.

Microsoft: The Next Catalyst Could Be Azure
Wall Street analysts believe Microsoft's high AI capital expenditures this year have not yielded significant returns, making investors cautious. While Microsoft's cloud business was expected to benefit from its ties to OpenAI, growth has slowed for three consecutive quarters. However, analysts are generally optimistic about Microsoft's prospects in 2025. Goldman Sachs believes the next potential catalyst for Microsoft could be accelerated growth in Azure starting in the first quarter of 2025. Microsoft's new platform, Azure AI Foundry (preview), the upcoming AI Agent Service, and Copilot Studio, which is expected to be adopted by over 100,000 organizations, are driving customer adoption of Azure AI. Usage of Azure OpenAI Services has doubled in six months, indicating that customers are finding real-world applications for these technologies.

Jefferies analysts also noted: "Microsoft is a primary beneficiary of (generative) AI and will gain from infrastructure (Azure AI) and application (various Copilots) opportunities." Beyond cloud computing, Microsoft is also making strides in AI hardware. This year, Microsoft launched its first Copilot+ AI PCs and plans to collaborate with more manufacturers to release additional Copilot+ AI PCs in 2025.

Apple and Meta Bet on AI Hardware
Apple and Meta are also exploring new opportunities in AI hardware. Apple released new iPhones this year equipped with Apple Intelligence. WedBush analyst Dan Ives claims Apple Intelligence will help Apple reach a $4 trillion market capitalization by early 2025. Ives said: "Apple could sell over 240 million iPhones in fiscal 2025, marking the highest iPhone sales year in history." However, BTIG warned investors to remain cautious about Apple's stock until January 2025. Meta, meanwhile, launched AI smart glasses, though analysts are skeptical about their potential benefits.

So far, Reality Labs, the division responsible for all of Meta's glasses and virtual reality devices (like Quest), has been operating at a loss, with a $13 billion deficit in the first three quarters of 2024 and only about $1 billion in revenue. Analysts believe AI agents and applications will be the key to Meta's revenue growth in 2025.

Despite the U.S. stock indices repeatedly hitting record highs this year, analysts are warning of potential bubble risks.
Bank of America cautioned that the U.S. stock market may have reached a tipping point and could face a double whammy next year from an AI bubble and the consequences of Trump's policies. AI shares similarities with the internet in the late 1990s, and a bubble burst is only a matter of time.
Goldman Sachs predicted in its latest report that 2025 will be the worst year for the relative performance of the "Magnificent Seven" stocks since 2017.
Steven Jon Kaplan, CEO of the True Contrarian blog and newsletter, believes the tech bubble in U.S. stocks has grown even larger. Kaplan, who accurately predicted several tech sell-offs in recent years—including advising investors to buy during the pandemic sell-off in 2020—said: "Most hedge funds now follow an algorithm where if stocks like QQQ (Invesco QQQ Trust) drop 20%, those trillion-dollar funds will sell simultaneously."