As Intel and Microsoft push AI PCs, Micron could be a big winner.

chip giant intel and software titans microsoft Each is working to bring artificial intelligence to personal computers. Intel is confident that integrating dedicated AI hardware into its CPUs will give it an edge and increase demand for PCs, while Microsoft is beefing up its Windows operating system by launching an AI-based assistant.

The two strategies do not fully converge. Intel’s latest Meteor Lake laptop CPUs include powerful AI hardware that can enable AI features in third-party software, but not enough to power features like OpenAI’s GPT-4, the advanced large-scale language model that Microsoft uses in its Copilot AI assistant. has almost no horsepower. So Microsoft’s AI assistant calls cloud services for processing.

Two disadvantages of using cloud services are latency and cost. Cloud-based AI models take time to respond, especially if they are overloaded with requests. This latency affects user experience. It’s also expensive. Microsoft offers the basic version of Copilot for free, but Copilot Pro costs $20 per user per month for users who already have a cloud-based Microsoft 365 subscription.

Software developers looking to integrate AI into their products have two options: Either run the AI ​​processing locally and incur no ongoing costs, or call a cloud service and pass those costs on to the user. For free software, the first option is the only viable option.

Intel will likely include AI hardware in most of its future CPUs, and that hardware will likely become more powerful over time. However, AI PCs require significantly more memory to run advanced AI models locally. That’s where memory chip manufacturers come in. micron (M.U. -2.74%) Come in.

either way is a winner

The memory semiconductor industry is emerging from a severe recession in which demand and prices are plummeting. One thing that will help firm up prices is surging demand for the ultra-fast memory used by AI accelerators running in cloud data centers. The supply of standard memory chips for PCs will shrink as Micron and its competitors prioritize this memory.

This will increase prices in the short term, but the requirements for AI PCs will increase demand in the long term. Laptops with 8GB of random-access memory (RAM) are still common today, but Microsoft has stated that all Windows PCs must have at least 16GB of RAM to be called AI PCs.

Even 16GB won’t get you very far. Large language models powerful enough to be useful as AI assistants may require significantly more memory. Llama 2 family of AI models created by metaFor example, you’ll need at least 30GB of RAM for the beefiest variant, and 320GB for the most powerful version.

Intel’s efforts to include AI hardware in CPUs will eventually make it possible to run powerful AI assistants on PCs without the need for cloud-based AI services. This would require a massive increase in the amount of RAM inside a typical PC, which would be a huge boost for Micron.

Even though AI PCs will take years to gain traction, Micron can now benefit from strong demand for memory chips used in powerful AI accelerators that train and run AI models in the cloud.

One thing to remember

Micron is an obvious beneficiary of the AI ​​revolution, but it’s important to remember that memory chips are a commodity and their prices are a function of supply and demand. Forecasts for the AI ​​accelerator market size are very optimistic. AMDFor example, we expect $400 billion worth of AI accelerators to be sold by 2027.

It’s these kinds of wild predictions that are planting the seeds for the next recession in the memory chip industry. All it takes to produce memory chips whose prices are prohibitively high as supply increases to meet expected demand is for these predictions to turn out to be a little too optimistic.

The memory chip market size is likely to grow as the demand for AI in both the cloud and personal devices increases, increasing memory chip content in PCs and servers. However, pricing, and therefore Micron’s profitability, depends on supply versus demand.

Micron is well-positioned to benefit from the growth of AI, but that doesn’t necessarily mean its profits will be permanently higher. This is something investors must understand before buying stocks.