<div id="readability-page-1">Nvidia (NVDA 1.74%) has dominated the artificial intelligence (AI) computing hardware game for the entirety of the AI arms race, but there's one company aiming to change that: Broadcom (AVGO 3.50%). While Nvidia's graphics processing units (GPUs) are best-in-class, Broadcom is looking to challenge Nvidia with something different: custom AI accelerators, which it calls XPUs. Is this the hardware that can dethrone Nvidia as the best AI investment in the market? Or is it just a management team driving hype around its innovations? <h2>XPUs by Broadcom will unlock a new level of AI innovation</h2> GPUs are valued because they can process multiple calculations in parallel, which makes them fantastic for arduous workloads like AI training and inference. Furthermore, GPUs can be connected in clusters to amplify this effect across multiple GPUs, allowing the AI giants to scale this to data centers that contain over 100,000 GPUs. This kind of raw training power is incredible, but GPUs are designed to be flexible and usable for a wide variety of tasks. The XPU is different because it is designed to process only one type of workload in a specific manner. This makes it less useful for other computing tasks, but perfect for AI training when the workload is configured in a specific way. Fewer wasted resources go to unused capabilities, making XPUs more powerful than GPUs, which is why Broadcom expects so much growth. In 2024, AI-related revenue for Broadcom totaled $12.2 billion. However, Broadcom projects that the market for its XPUs and other AI-related hardware to service GPUs will be between $60 billion and $90 billion by 2027. Furthermore, that market is only being driven by three customers. This type of hardware is starting to become more popular, and four more clients are working on getting their XPU design up and running. This will further expand Broadcom's monstrous market projection, making it an intriguing stock for investors to watch. But does this spell the end of Nvidia? <h2>Nvidia's business isn't in too much danger right now</h2> One analyst on Broadcom's Q1 FY 2025 (ending Feb. 2) conference call had the same question, as he inquired whether an inference-heavy market favors GPUs or XPUs. Broadcom CEO Hock Tan noted that the massive market that's expected to emerge over the next few years still includes GPUs, and it isn't a perfect split. This is important to keep in mind, as GPUs still have a use in AI training and inference. Not everything these companies do can be properly configured to run on an XPU, so Nvidia GPUs will still be a huge part of AI expenditures from the hyperscalers. Furthermore, another large segment of GPU usage is in data centers rented out to clients in cloud computing. Many of these clients already know how to run a GPU workload efficiently, and switching to an XPU would be very difficult. So, I'd say Nvidia's growth will still be fine over the next few years; just look for Broadcom to eat into some of Nvidia's market share as XPUs become more popular. But, at today's prices, both stocks look like pretty strong buys. <img src="https://imageproxy.pbkrs.com/https://media.ycharts.com/charts/20fff2cdba04912a2d7d7ff6b9b4e760.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" width="720" height="456" original-src="https://imageproxy.pbkrs.com/https://media.ycharts.com/charts/20fff2cdba04912a2d7d7ff6b9b4e760.png"/> AVGO PE Ratio (Forward) data by YCharts With each company expecting huge growth over the next few years, paying less than 30 times forward earnings seems like a great price, especially considering where these stocks have traded over the past few years. I think the pessimism has already been baked into Nvidia's stock, and the optimism for Broadcom's growth has yet to emerge. As a result, each stock looks like an excellent buy right now, and investors would be smart to scoop up shares while they're still on sale. </div>

NVDA

AVGO

Broadcom is positioning itself to challenge Nvidia's dominance in AI computing with its custom AI accelerators, known as XPUs. While Nvidia's GPUs are versatile, Broadcom's XPUs are tailored for specific AI workloads, potentially making them more efficient. Broadcom anticipates its AI-related revenue to grow significantly, projecting a market of $60-$90 billion by 2027. Despite this, Nvidia's GPUs will remain essential for many applications, and both companies are seen as strong investment opportunities at current prices.

- Nvidia has been the leader in AI computing hardware, but Broadcom is introducing custom AI accelerators called XPUs to challenge this dominance.  
- Broadcom's XPUs are designed for specific workloads, potentially offering greater efficiency than Nvidia's versatile GPUs, with projected AI-related revenue growth from $12.2 billion in 2024 to between $60 billion and $90 billion by 2027.  
- Despite this competition, Nvidia's business remains stable, as GPUs will continue to play a significant role in AI applications, making both Nvidia and Broadcom attractive investment opportunities.

Is Broadcom Primed to Take Over Nvidia's Market Share? 1 Jaw-Dropping Projection That Investors Must Be Aware Of.