Investors in NVIDIA (NVDA.US) and all AI-related Stocks have high expectations for CEO Jensen Huang's CES speech on Monday local time, hoping that his latest remarks will significantly boost the recently slightly sluggish "AI faith," thus driving a renewed wave of investment in AI to sweep through the financial markets, leading to a new upward trend in the stock price of this AI chip leader while also pushing all stocks related to artificial intelligence into a new rising curve.

Since November of last year, NVIDIA's stock price seems to have remained stable; in contrast, for most of 2024 and throughout 2023, it was in a "skyrocketing mode". Investors in NVIDIA are looking forward to Jensen Huang's speech at CES triggering a new breakthrough in the stock price, while "AI believers" hope that Huang's remarks will also spark a new uptrend in prices for all AI-related stocks, including AMD, Broadcom, and the PHLX Semiconductor Index ETF.

The latest research advancements on AI Chips, as well as edge artificial intelligence, are expected to be the core focus of the 2025 International Consumer Electronics Show (CES), particularly discussing how AI can more perfectly integrate into edge Consumer Electronics and how it can deeply penetrate human daily life, especially the groundbreaking new products in edge AI field such as "AI Humanoid Robot" and the development progress of core robotics technology, which are drawing global investors' and tech industry fans' attention.

Therefore, in the upcoming CES speech that is drawing global investors' attention, the market's core focus will be on the demand outlook for NVIDIA's Blackwell architecture AI GPU and the latest advancements in NVIDIA's AI-related technologies in the humanoid robot field. CEO Jensen Huang may reveal the latest demand outlook for Blackwell and provide more information about the upcoming Blackwell Ultra and Rubin architectures, and may also showcase the latest humanoid robot strategy and the Jetson Thor platform designed specifically for humanoid robots at CES 2025.

Jensen Huang will make a significant appearance at the Las Vegas International Consumer Electronics Show (CES), which is attracting attention from Wall Street investment institutions and global investors. The NVIDIA CEO typically uses this major global tech event to showcase consumer electronic devices using its chip products, as well as forward-looking news about the next generation of AI GPU architecture.

Investors will focus on any outlook-related comments regarding the Blackwell series AI GPU on Monday local time; this AI chip architecture is seen by the financial markets as NVIDIA's next core performance growth catalyst, as well as the strongest propellant driving NVIDIA's performance to continue to exceed expectations. According to information disclosed by NVIDIA executives at the last Earnings Conference, despite very strong demand, the Blackwell AI GPU still faces supply constraints, partly due to manufacturing challenges slowing its rollout, and the capacity limitations of NVIDIA's exclusive chip Contract Manufacturers, Taiwan Semiconductor for AI Chips.

The Blackwell architecture AI GPU series products are undoubtedly the "performance ceiling" in the current AI computing infrastructure field, and the barrier of the NVIDIA CUDA ecosystem forces technology companies to continuously spend heavily on purchasing NVIDIA AI GPUs and a broader AI infrastructure. The demand from tech giants like Microsoft, Amazon, and Google, as well as the strong demand from software companies like Salesforce and ServiceNow, explains why the market expects Blackwell to drive persistent explosive growth in NVIDIA's performance.

As a benchmark, Blackwell's performance is much stronger than the previous generation "performance ceiling" Hopper. In the MLPerf Training benchmark tests, Blackwell's performance in the GPT-3 pre-training task was significantly improved by 2 times per GPU compared to Hopper. This means that with the same number of GPUs, model training can be completed faster using Blackwell. For the Llama 2 70B model's LoRA fine-tuning task, Blackwell's performance per GPU was improved by 2.2 times compared to Hopper, indicating higher efficiency when handling specific high-load AI tasks.

Matt Chopa, a portfolio manager from Franklin Templeton's equity business, stated: "Analysts expect Blackwell's demand to remain very strong. This may cause the financial markets to refocus on NVIDIA's ultimate long-term opportunities, thereby pushing the stock price into the next breakout."

Even before Jensen Huang's CES speech, NVIDIA's investors had plenty of reasons to remain optimistic about the stock price trends. Over the past six months, Huang's statements about AI chip demand have significantly boosted the company's stock price. In October, he emphasized that the market demand for Blackwell was "crazy," and in November, he stated that due to "very strong" demand, the first batch of Blackwell would be delivered in large quantities within this fiscal quarter (the three months ending at the end of January).

However, under the continued pressure of high interest rate expectations brought by the hawkish Federal Reserve, the stock experienced a rare "monthly decline" in December 2024 since 2023, but still rose by 171% in 2024. NVIDIA, with its high weighting, also became the largest single driving factor for the overall increase in the S&P 500 Index in 2024.

Since the beginning of this year, NVIDIA's stock price has risen nearly 8%, and the current trading price is only 3% lower than the historical highest closing price. Huang's latest speech could even drive NVIDIA's stock price to a new historical high. NVIDIA currently has a market cap of approximately $3.54 trillion, second only to Apple.

The latest earnings reports did not lead to a sustained surge in NVIDIA's stock price, and the market is looking forward to new catalysts.

After NVIDIA released its latest Earnings Reports on November 20, Eastern Time, the company's stock price briefly hit an all-time high, but subsequently plummeted for a while. The company’s performance outlook did not impress Wall Street, which has become accustomed to NVIDIA's performance outlook significantly exceeding average expectations, making the threshold for NVIDIA's outperforming expectations increasingly high.

As NVIDIA's stock price remained almost stagnant, excitement over AI spending spread to other areas of the semiconductor Industry. For example, in the field of AI ASIC chips, which is currently NVIDIA's strongest AI infrastructure competitor.

Broadcom (AVGO.US), the strongest player in AI ASICs, saw its stock price soar nearly 30% in the past few weeks, entering the 'trillion-dollar stock market cap club'. Previously, the company predicted an unprecedented boom in its AI core hardware components market designed for Datacenter operators. Another major player in AI ASIC chips, Marvell Technology (MRVL.US), has seen its stock price rise more than 20% since its Earnings Reports showed that customized AI chip demand significantly exceeded expectations.

However, from the perspective of major Wall Street firms such as Morgan Stanley, there is no so-called 'complete substitution' between AI GPUs and AI ASICs; these two distinctly different AI hardware paths will 'coexist harmoniously' for a long time. Combining ASICs with GPUs will enable AI ASICs to be more focused on normalized AI tasks or structurally fixed AI applications, such as common operations in deep neural networks like matrix operations and convolution operations, while NVIDIA's AI GPUs will focus on much more complex and computationally intensive AI tasks, such as self-attention mechanisms, and neural networks composed of generators and discriminators in Generative Adversarial Networks (GANs), as well as broader and larger general computing tasks. GPUs have very strong floating-point calculation capabilities and high memory bandwidth, making them suitable for tasks that require large amounts of data and complex algorithms.

The stock analyst team at Morgan Stanley, led by Joseph Moore, likened the rise of AI-related stocks such as Broadcom to 'a brief transfer of NVIDIA's AI wealth'. After Broadcom's extraordinary Earnings Reports, NVIDIA's stock price fell for four consecutive trading days, with a Market Cap shrinkage of over 200 billion dollars. However, NVIDIA remains Morgan Stanley's 'preferred stock', as analysts believe this chip giant will still dominate the market share of AI infrastructure this year. They also expect Jensen Huang's keynote speech at CES to be a 'positive catalytic big event'.

'The message should be the same - the demand for Blackwell is very strong, but the supply is limited,' they wrote in a recent research report. 'By mid-year, we still believe that the market focus will remain on Blackwell, which will be the core driver of revenue in the second half of the year.'

The analyst team at Bank of America recently released the "Preferred Chip Stocks List" for the U.S. stock market in 2025, which includes the "three giants of AI chips" that have seen explosive investment interest in the U.S. stock market since 2023—NVIDIA, Broadcom, and Marvell Technology, all of which are featured on Bank of America's 2025 "Preferred Chip Stocks List." The list also includes semiconductor equipment giant Lam Research (LRCX.US), leading automotive chip manufacturer ON Semiconductor (ON.US), and one of the leaders in EDA Software, Cadence Design Systems (CDNS.US). Bank of America sets a mid-term target price of up to $190 for NVIDIA. As of last Friday's U.S. stock market close, NVIDIA's stock price was $144.470.

For NVIDIA investors, this can be considered a "do-or-die" moment.

Jordan Klein, a technology industry expert from Mizuho Securities, believes that any hot topics about AI at CES and Jensen Huang's AI-themed keynote speech will be a critical test of investor sentiment and risk appetite in the Technology industry.

He wrote that if NVIDIA's stock price falls or stagnates in the days following Jensen Huang's speech, "In my view, this would have a moderate negative impact on the Global Equity markets in January." He also mentioned that since the earnings season for technology companies will not begin until later this month, "before that, investors who favor tech stocks have almost no positive basis to assess fundamentals and prospects."

After Jensen Huang's speech, the fluctuations in NVIDIA's stock price could be substantial, as its price surge has heightened concerns about valuations. The stock's expected PE ratio based on revenue outlook is about 18x, making it one of the 10 stocks with the highest expected PE ratios in the NASDAQ 100 Index. The expected PE ratio based on EPS is also close to 34x, while the chip benchmark—the PHLX Semiconductor Index—has an expected PE ratio of about 24x.

The chief investment strategist team from John Hancock Investments remains optimistic about NVIDIA and other large-cap U.S. tech stocks but expects "more volatility" in the future. "At some point, in 2025, there will be a pivotal moment of 'show me' when this AI narrative logic will have to be reaffirmed. Valuation is clearly an issue, and it's important to note how high the market's valuation can be despite the high-speed growth in performance, even if the tailwinds of AI remain intact."

In addition to Blackwell, the market is also focused on humanoid robots.

We are likely to hear more from Jensen Huang regarding the most detailed market supply and demand news for the company’s Blackwell architecture AI GPU, which will begin its first supply at the end of 2024 and may start mass shipments early this year.

CES 2025 may become a promising start for the increasingly mature generative AI software ecosystem combined with humanoid robots. Wall Street analysts generally expect NVIDIA to showcase its latest humanoid robot strategy and the Jetson Thor platform designed for humanoid robots at the 2025 CES, which will be closely related to NVIDIA's "Embodied AI" strategy.

When the deployment scale of AI Chips in datacenter servers handling massive parallel computational demands meets the baseline computing power requirements of Global enterprises and individuals, as well as the enormous computational support needed for the iterative development of AI large models, according to technology development trends, in application terminals including Consumer Electronics such as Smart Phones, PCs, and humanoid Robots, as well as electric vehicle software systems and industrial production applications, the increasingly powerful AI large models with inference capabilities similar to GPT-4o and o1 are bound to be integrated into these terminal devices and be able to call on more powerful cloud-based AI computing power for instant collaboration, which is referred to as edge AI.