Big Tech’s appetite for AI infrastructure isn’t being sated anytime soon, and Broadcom Inc. (NASDAQ:AVGO) is proof of that. On Wednesday, the custom AI chipmaker said it now expects AI chip revenue of about $115 billion in the fiscal year ending October 2027, up from a prior forecast of over $100 billion. Broadcom also expects this revenue to double to roughly $230 billion in fiscal 2028.
Broadcom’s revenue outlook is evidence that hyperscalers no longer want costly processors only. Rather, they are also interested in suppliers such as Broadcom who provide both the custom chips and the networking components. Broadcom’s forecasts are therefore a reflection of continued spending by Big Tech on alternatives to NVIDIA Corporation (NASDAQ:NVDA)’s GPUs, as well as components that complement them.
In response to this forecast, KeyBanc analyst John Vinh reiterated an Overweight rating and $575.00 price target on Broadcom.
Hyperscalers Want Suppliers Like Broadcom
Hyperscalers are increasingly looking for suppliers that offer different types of processors, designed for different types of workloads. This is because they want more control over the economics of their AI infrastructure. The preference doesn’t just reduce their reliance on Nvidia, but also offers them the chance to reduce operational costs and build custom silicon.
Broadcom President and CEO Hock Tan has confirmed that the demand for both its custom AI accelerators and AI networking infrastructure continues to outpace supply. Particularly in its Q3 fiscal 2026 earnings report, Tan noted exploding demand driven by major frontier customers such as OpenAI, Meta, Google, and Anthropic.
Keybanc’s note highlights how Anthropic/ OpenAI would be Broadcom’s largest customers in fiscal 2028. This could offer Broadcom Inc. more visibility beyond its Google relationship. The expansion reflects how custom chips are increasingly becoming an important part of AI infrastructure.
However, competition within the segment is also becoming intense. Alphabet’s Google recently expanded its deal with Marvell, with analysts even pointing to AVGO losing share in Google’s TPU programs. While the deal doesn’t aim to displace Broadcom Inc., it does imply that the position isn’t guaranteed either.
Broadcom comes from a place of operational strength, with fiscal third-quarter revenue rising 86% year-over-year to about $29.6 billion and topping Wall Street expectations. However, its fourth quarter guidance of roughly $34.8 billion landed just under the $35 billion consensus estimate- a mixed guidance flagged in KeyBanc’s note.
Custom Chips Don’t Necessarily Threaten Nvidia’s GPU Dominance
The bull case for Nvidia in this scenario is that hyperscalers aren’t being seen abandoning it. Rather, they are adding custom silicon alongside GPUs. Reuters noted Nvidia holding 80% of the GPU market. AWS is deploying Trainium alongside millions of Nvidia GPUs, Microsoft is scaling both Nvidia Rubin and Maia, while Google is also offering TPUs and using Nvidia GPUs.
Moreover, Nvidia’s large developer ecosystem, CUDA software, and hardware performance are difficult to replicate. At the same time, it should be noted that hyperscalers also have the scale to justify spending on chips around specific workloads. This means they can also eventually route a large portion of workloads away from expensive general purpose GPUs.
Hedge Fund Analysis
According to Insider Monkey’s database, 170 hedge funds held positions in the stock at the end of the Q2, modestly down from 173 in the previous quarter. As for Nvidia, 285 hedge funds held positions in the stock at the end of the second quarter, up from 275 in the previous quarter.
Filings also reveal how Fisher Asset Management held about 15.1 million shares in Broadcom and 90.9 million shares in Nvidia at the end of the second quarter. AQR Capital Management increased its position in Broadcom by 9%, while simultaneously increasing its position in Nvidia by 18%.
Overall, Broadcom’s forecast doesn’t imply that the demand for Nvidia’s GPUS is slowing down. Rather, it suggests that hyperscalers are increasingly preferring both Nvidia and suppliers such as Broadcom. If AI infrastructure continues to expand at unprecedented rates, both companies are set to gain from it.
READ NEXT: NVIDIA (NVDA): What Foxconn and Super Micro Are Telling Us about the AI Boom and Snowflake (SNOW) Stock: AI Growth Is Real, But Is the Valuation Already Priced In?
