Source: Zhitong Finance.
The upcoming GTC 2025 conference - referred to by Wall Street as the "Woodstock Music Festival of AI" - is likely to become a key turning point for NVIDIA.
In the past two years, $NVIDIA (NVDA.US)$ there has been a dramatic turnaround from the best start to the worst start in the USA stock market. This highlights the drastic change in market sentiment, which has not been alleviated at all by NVIDIA's core position in the AI revolution.
The quick change in sentiment can be attributed to multiple adverse factors, leading investors to take profits or even exit the market entirely. The main factor is the market narrative of weakening economic resilience, putting growth stocks like NVIDIA at a critical turning point.
However, NVIDIA's management is well aware of this, so the upcoming GTC 2025 conference - referred to by Wall Street as the 'Woodstock Music Festival' of the AI world - is likely to become a key turning point for NVIDIA.
Highlights of the conference
From March 17 to 21 local time, NVIDIA will hold the GTC 2025 conference in San Jose, California, USA. This conference is expected to bring three major highlights:
1. New products, product update cycles, and full-stack AI ecosystem upgrades.
NVIDIA currently relies on two main GPU architectures to drive the growth of the Datacenter Business. The H100 GPU, powered by the Hopper architecture, achieved a 120% compound annual growth rate from 2023 to 2024, while the Blackwell architecture contributed $11 billion in revenue in the fourth quarter of fiscal year 2025, accounting for 28% of the total revenue of $39.3 billion.
In last month’s fourth quarter Earnings Reports conference call, Huang Renxun revealed updates on the Blackwell Ultra (B300 GPU or GB300 system) and the next-generation GPU architecture named Vera Rubin.
(Note: NVIDIA has transitioned to an annual release cycle for GPU architectures.)
Regarding the Blackwell Ultra (B300 GPU/GB300 system), supply chain news indicates that the price of the GB300 server system will be higher than that of the GB200 system by $3 million.
Given that the current cost of the GB200 NVL72 server is approximately $3 million, the price of the top configuration for the GB300 will be higher, which will ultimately translate into more revenue for NVIDIA. However, considering the yield issues with Blackwell, whether customers will choose the GB300 in large volumes may be a point of concern.
Additionally, the GTC conference agenda shows that NVIDIA will hold an event to announce updates to CUDA. This is the company’s proprietary GPU programming language, regarded as NVIDIA's core Software competitive advantage, and investors are expected to look forward to the content of this event.
2. Eliminate industry noise.
Traditionally, discussions about Datacenter computing have always revolved around NVIDIA and its two direct competitors - AMD (AMD.US) and Intel (INTC.US). Among these three, NVIDIA continues to dominate the market. However, over the past 8 to 12 months, large-scale Datacenter operators have increasingly leaned towards developing their own accelerator chips based on their custom requirements.
This has led to a shift in the narrative around NVIDIA's competition from AMD/Intel to commercial GPUs versus custom GPUs or XPUs. Broadcom (AVGO.US) has stood out as a potential winner in this changing narrative.
During the Earnings Reports conference call, Jensen Huang issued a warning: "The completion of chip design does not mean successful deployment."
This comment suggests that self-developed chips encounter practical obstacles. It is understood that during the procurement process of XPUs, these chips are first designed and tested at the sites of large-scale Datacenter operators. Only after successful testing will production Orders be granted to XPU manufacturers like Broadcom.
In light of the trend of large-scale Cloud Computing Service companies developing their own accelerator chips, NVIDIA has also strategically defended by emphasizing TCO (Total Cost of Ownership) advantages.
The company's management stated: "Our core goal in system building is to find the optimal TCO solution for our customers. It is noteworthy that every time we hold the GTC conference and launch a new architecture, we find that a large number of designs actually never entered the market, and some projects are even abandoned halfway or require re-investment in R&D. Therefore, I believe the market landscape has not fundamentally changed. While customers may see customized solutions as an opportunity, almost all customers are still using NVIDIA's solutions, even if they might attempt some custom development."
However, judging by market reactions, it seems that NVIDIA's statements have not convinced investors. With companies like Broadcom entering the fray, NVIDIA urgently needs to demonstrate its response strategy and update its outlook on the GPU market at the GTC conference.
3. Breaking through the trillion-dollar market ceiling.
At last year's Goldman Sachs Communacopia conference, Jensen Huang stated that he expects there will be a trillion-dollar general Datacenters globally upgraded with accelerated computing resources, such as NVIDIA's GPUs, to handle AI workloads. The company's recently announced fiscal year 2025 full-year performance shows a consolidated revenue growth of 114%, reaching $130.5 billion, indicating that NVIDIA's penetration rate in the trillion-dollar TAM predicted by Jensen Huang is approximately 13-14%.
The question is whether the products and updates from next week's GTC event can expand NVIDIA's TAM to over a trillion dollars. Can Blackwell Ultra do for NVIDIA what Hopper has done over the past two years?
The emergence of DeepSeek has transformed the narrative of AI capital expenditures into a discussion about efficiency and higher ROI, which will require NVIDIA to reshape its product value proposition from a cost optimization perspective.
Currently, NVIDIA's management has begun to push the narrative about how NVIDIA GPUs can bring significant TCO savings, which is also the cost Indicators used by the aforementioned hyperscale Datacenter operators and Cloud companies to assess the costs of owning and running Cloud Computing Services.
In last month's fourth quarter Earnings Reports conference call, Jensen Huang also mentioned TCO multiple times. Therefore, Blackwell Ultra will help the company improve the narrative on how to reduce TCO for its top customers, which will be a key focus of the market.
Low valuation.
Finally, from a risk/return perspective, NVIDIA's Stocks seem quite attractive after a round of sell-off.
For growth stocks, NVIDIA's current PEG ratio is only 0.73, which means the company's valuation has a significant discount compared to expected growth. Currently, the market generally expects the company's EPS for the fiscal year 2026 to grow by 50%, reaching $4.49 per share, with revenue growing by 56%, reaching $200 billion.
Although its GAAP valuation multiple has reached 26 times, this valuation is also about 58% lower than NVIDIA's five-year average.
However, investors should also be aware of the two major risks NVIDIA may face: regulatory scrutiny and export control measures, including global tariff fluctuations, which will still impact NVIDIA's stock price in the medium term.
Summary
Next week's GTC 2025 event will undoubtedly be an important catalyst for NVIDIA's stock price, and given NVIDIA's stock performance this year, the company indeed needs a new narrative direction for growth.
After a significant pullback in the stock price, NVIDIA is clearly attractive, but whether investors are ultimately willing to bottom fish may still depend on how much confidence Jensen Huang brings to investors at the GTC conference.
Editor/jayden
