Both Nvidia (NVDA +0.87%) and Broadcom (AVGO +0.29%) have been big beneficiaries of the huge investments in artificial intelligence (AI) hardware, which explains why the share prices of both companies shot up impressively over the past couple of years.
While Nvidia stock jumped more than 280% in the past two years, Broadcom recorded healthy gains of 160% during the same period. However, both AI stocks witnessed a substantial pullback in 2025 despite reporting solid growth in revenue and earnings. Nvidia stock retreated close to 18% this year, while Broadcom has recorded a sharper drop of almost 28%.
The decline in their share prices can be attributed to the negative stock market sentiment this year arising out of the economic uncertainty created by the tariff-induced trade war. The probability of a U.S. recession has also been going up of late, which explains why investors are becoming risk-averse and are booking profits in stocks that have performed strongly in recent years.
NASDAQ: NVDA
Key Data Points
However, the pullback in shares of Nvidia and Broadcom could actually be a good thing for savvy investors looking to add potential long-term winners to their portfolios. That's because the AI market is currently in its early phases of growth and is expected to show an annual growth rate of 36% through 2030.
But if you have to choose one of these two AI stocks right now, which one should it be? Let's find out.
The case for Nvidia
Nvidia kicked off the AI revolution three years ago with its powerful data center graphics processing units (GPUs) that were used by OpenAI to train its highly popular chatbot, ChatGPT. The chip designer became become the go-to supplier of AI GPUs for tech giants and governments across the globe, which resulted in a remarkable increase in the company's revenue and earnings.
NVDA Revenue (TTM) data by YCharts
Nvidia established a solid grip over the data center GPU market, controlling an astonishing 92% of this space in 2024, as per a third-party estimate. Nvidia's recent results indicate that it is unlikely to yield much ground in the data center GPU market as its latest products have become a runaway hit. Specifically, the company sold $11 billion worth of its latest generation of Blackwell AI GPUs in the previous quarter.
Nvidia says that Blackwell "is the fastest product ramp in our company's history, unprecedented in its speed and scale." The initial success of the Blackwell processors can be gauged from the fact that rival Advance Micro Devices sold over $5 billion worth of AI GPUs last year. Looking ahead, the robust demand for Nvidia's AI GPUs seems here to stay as the company is witnessing strong demand for both AI training and inference applications.
The company seems to have benefited from the recent breakthrough achieved by the arrival of cost-efficient reasoning models such as DeepSeek's R1, which led to an increase in demand for computing power. On its February earnings conference call, Nvidia management pointed out:
Our inference demand is accelerating, driven by test time scaling and new reasoning models like OpenAI's o3, DeepSeek-R1, and Grok 3. Long-thinking reasoning AI can require 100x more compute per task compared to one-shot inferences.
Blackwell was architected for reasoning AI inference. Blackwell supercharges reasoning AI models with up to 25x higher token throughput and 20x lower cost versus Hopper 100.
Not surprisingly, Nvidia's rivals have been scrambling to accelerate the launch of new chips so that they can keep up. But that's easier said than done, as the chipmaker built a solid supply chain ecosystem and has a robust developer community that gives it an advantage over peers. Nvidia management remarked on the conference call that its "programmable architecture accelerates every AI model and over 4,400 applications, ensuring large infrastructure investments against obsolescence in rapidly evolving markets."
As a result, the company has been able to achieve a 200x reduction in AI inference costs over the past couple of years. So, it won't be surprising to see Nvidia remaining the top player in the AI chip market in the long run, which is expected to generate a whopping $360 billion in annual revenue in 2030. Nvidia's data center revenue grew 142% in the latest fiscal year to a record $115 billion, and the size of the end-market opportunity suggests that there is still a lot of room for growth here.
All this explains why Nvidia's revenue growth estimates for the next three fiscal years have seen upward revisions, with the company expected to show healthy double-digit growth even after the remarkable pace at which it has been growing in recent quarters.
NVDA Revenue Estimates for Current Fiscal Year data by YCharts
Nvidia, therefore, should continue to remain a top AI stock going forward. However, don't be surprised to see Broadcom follow in its footsteps.
The case for Broadcom
Broadcom doesn't compete directly with Nvidia in AI chips, as it sells application-specific integrated circuits (ASICs). These are custom processors that are designed to perform specific tasks, unlike GPUs, which are meant for general computing tasks. The specialized nature of the custom AI processors means that they can be more efficient and powerful while handling AI workloads, which is precisely the reason why Broadcom is seeing outstanding growth in its AI revenue.
NASDAQ: AVGO
Key Data Points
Broadcom's AI revenue more than tripled in the recently concluded fiscal year 2024 to $12.2 billion. However, a closer look at management's comments on the company's March earnings conference call suggests that its AI revenue could multiply impressively in the coming years. Broadcom's custom AI processors and networking chips are currently used by three hyperscale cloud customers.
The company sees a potential annual revenue opportunity worth $60 billion to $90 billion from these three AI customers over the next three fiscal years. However, Broadcom has an additional four hyperscale customers that are utilizing its designs to manufacture custom AI chips. Two of these customers are on track to tape out their chips in 2025, which is the final development stage before the chips are sent for manufacturing.
The other two hyperscale customers that Broadcom has added of late are going "to develop custom accelerators to train their next-generation frontier models." All this indicates that Broadcom's eventual AI-related revenue opportunity could be much larger than what the company is estimating. So, there is a solid chance that Broadcom may be able to outgrow Wall Street's growth expectations in the current and the next couple of fiscal years.
AVGO Revenue Estimates for Current Fiscal Year data by YCharts
A big reason why Broadcom's growth could be better than consensus estimates is because it is the leading player in the ASIC space, with an estimated market share of 55% to 60%. Broadcom's recent customer wins suggest that it could hold on to its impressive share in the future, which is why there is a strong likelihood that it will be able to significantly multiply its AI revenue in the future.
The verdict
It is clear that both Nvidia and Broadcom are dominant players in their respective AI semiconductor niches, which is why they are expected to keep growing at healthy rates going forward. What's more, their valuations tell us that investors cannot go wrong with either stock.
Nvidia and Broadcom sport almost similar forward earnings multiples.
NVDA PE Ratio (Forward) data by YCharts
Investors looking to buy an AI stock right now can consider buying either Nvidia or Broadcom. In fact, they may even consider adding both these names to their portfolios as they are on track to capitalize on different but lucrative niches of the AI semiconductor market that should help them sustain their outstanding growth for years to come.
