Advanced micro devices (AMD -4.60%) recently unveiled its latest artificial intelligence (AI) chip that aims to remedy the situation Nvidia‘S (NASDAQ:NVDA) huge market share. While AMD, as the company is commonly known, has seen its AI sales skyrocket this year, it’s still only a fraction of the revenue generated by its rival’s graphics processing units (GPUs).

The question is: Can this new chip help AMD become the next big AI winner?

An AI winner or an AI afterthought?

AMD has steadily increased its data center GPU sales this year, from an initial forecast of more than $3.5 billion to a recent $4.5 billion. However, that pales in comparison to the $26.3 billion in data center revenue that Nvidia just generated last quarter.

The semiconductor company hopes its latest chip, called Instinct MI325X, will help close the large GPU market gap between the two companies. According to AMD, the new chip offers industry-leading memory capacity and bandwidth, as well as more processing power than Nvidia’s H200 chip. However, it’s worth noting that the comparison is with Nvidia’s Hopper architecture and not the latest Blackwell architecture. For its part, Nvidia claims that Blackwell can have 30 times higher inference performance than Hopper.

AMD said the new chip will begin production in the fourth quarter and be ready for mass shipping in early 2025. The chips would be available across a wide range of platform providers including Dell, Super microcomputerAnd Hewlett Packard Enterpriseamong other things. Like Nvidia, AMD is shifting its AI chip development cycle to about once a year.

Image source: Getty Images.

The company also introduced some networking solutions, including a new NIC (network interface card) and DPU (data processing unit). NICs are needed to connect to a network, while DPUs handle data transfer, compression and storage. It also introduced a new range of CPUs (central processing units) called EPYC 5th Gen, said to be designed to accelerate data center, cloud and AI workloads.

Meanwhile, AMD announced that it is adding some powerful new features and capabilities to its open ROCm software stack. While AMD has produced powerful GPUs in the past, Nvidia’s CUDA software has long been a major asset to the company, as developers have long been trained to program GPUs through its platform. This has created a huge competitive advantage for Nvidia that AMD and others have struggled to compete against.

AMD has not announced any new customers for its new products, but representatives from alphabet, Metaplatforms, Microsoft, and OpenAI discussed how they use AMD at the event. Meta noted that it is “working on multiple training workloads with AMD.”

Is it time to buy AMD stock?

There was no indication in AMD’s announcement that the company will acquire significant shares of Nvidia in the near future. However, the AI ​​infrastructure market is growing so quickly and Nvidia has limited supply capacity, so there is still a good opportunity for AMD to expand its AI chip business. Additionally, companies will want to maintain Nvidia’s honesty and have at least a second GPU supplier so that Nvidia doesn’t become a complete monopoly.

There have also been some comments from customers suggesting that AMD’s GPUs are receiving more critical workloads than in the past. Combined with the strong overall GPU market, the company should continue to deliver solid AI revenue growth, just off a much lower base than Nvidia. And at the end of the day, AMD doesn’t have to take a ton of share from Nvidia to be an AI winner. Going from a 4% market share to 7% in a rapidly expanding market would still be a lot of growth.

From a valuation perspective, AMD is currently trading at a forward price-to-earnings (P/E) ratio of around 30, based on analyst estimates for 2025. That’s a fair valuation given the potential growth the company has ahead of it through AI.

AMD PE Ratio (Forward 1y) data from YCharts

AMD probably won’t be the next Nvidia, but it’s a solid secondary option in this space for investors looking to diversify their AI stock holdings. As long as the market for AI infrastructure grows strongly, the stock should perform solidly.