One of the pillars of the artificial intelligence (AI) boom is the semiconductor. And when it comes to high-performance graphics processing units (GPUs), "Magnificent Seven" member Nvidia is at the forefront of the discussion. But even as demand is soaring for the company's A100 and H100 chips, smart investors realize that there are other opportunities alongside Nvidia.
Considering that stock is up 200% over the last year, it might be worth checking out the competition and looking for a better value. Let's dig into three of Nvidia's primary competitors and assess why each company could be a good buy right now.
1. Intel
In early April, Intel (INTC 2.43%) made waves with an announcement about its new AI chip, the Gaudi 3. According to Intel, the Gaudi 3 has 1.5 times faster training and inferencing power over Nvidia's H100. Moreover, the chip uses 40% less power than Nvidia GPUs, offering customers nice savings.
Although Nvidia holds an estimated 80% share of the AI chip market, I'm cautiously optimistic that Intel can eat away at its rival's lead. Providers of IT architecture solutions -- including Dell Technologies, Hewlett Packard Enterprise, Lenovo, and Super Micro Computer -- will all be designing systems that use the Gaudi 3. And Intel announced that early Gaudi 3 customers and partners include IBM and SAP.
Investors got a nice surprise during the company's first-quarter 2024 earnings report when CEO Pat Gelsinger said he now expects over $500 million in accelerated revenue in the second half of the year following the release of Gaudi 3.
Although Nvidia is still the 800-pound gorilla in AI semiconductors, I see Intel's latest product release as an encouraging sign and am impressed by the initial success of Gaudi 3. With AI applications constantly evolving, I think Intel has a good chance to compete with Nvidia at a high level in the long run.
Image Source: Getty Images
2. Advanced Micro Devices
The second company on my list is Nvidia's closest competitor, Advanced Micro Devices (AMD 1.85%). As with Intel, Nvidia has a big lead on AMD. But this is not all good news for Nvidia, whose management has reminded investors that demand is outpacing supply, thereby creating a bottleneck for new business.
This is an opportunity for the competition to step in and take advantage. In late December, AMD announced its release of the MI300X accelerator. To date, this is the company's closest competing chip to what Nvidia offers.
While it's still early, the MI300X launch has been encouraging so far. During the fourth quarter, the company's data center GPU business outperformed expectations by $400 million thanks to accelerated demand for the MI300X. Moreover, management believes that the addressable market for such chips will reach $400 billion by 2027.
The combination of early success for the MI300X and an enormous market opportunity in data center GPUs gives AMD the opportunity to gain some momentum on Nvidia and become a more prominent player in the chip space over the long run.
3. Meta Platforms
I'll admit that Meta Platforms (META -1.73%) might look like the most unconventional company on my list. Indeed, today the company is largely an advertising leader thanks to its large Internet surface area covered by applications such as Facebook, Instagram, and WhatsApp.
However, as competition in social media heats up from TikTok, YouTube, Pinterest, and others, Meta is demonstrating that its impressive cash pile is being used for lucrative new opportunities. During the company's first-quarter earnings call, investors learned that management expects between $35 billion and $40 billion this year in capital expenditures. This is a noticeable increase from prior guidance of $30 billion to $37 billion.
The reason behind the increased capex is Meta's efforts to build its own chips internally. The company's Meta Training and Inference Accelerator (MTIA) chip could be a gold mine. First, a successful deployment of the MTIA would allow Meta to move away from its current infrastructure using Nvidia's H100 chips. This could lead to massive cost savings in the long run because the company would have its own customer chips and be less reliant on Nvidia over all.
The more subtle reason MTIA could unlock gains for Meta is that it provides a path for Meta to internalize as much of its tech stack as possible. The company is harvesting loads of data from its various social media platforms, so integrating its own chips into this workload could lead to more efficient operations as Meta seeks to continuously hone its targeted advertising capabilities.
The MTIA accelerator could be an enormous differentiator for Meta compared to peers, and I think it represents an entirely new source of growth for the company.
