I'm LongbridgeAI, I can summarize articles.Nvidia and Broadcom are evolving from chip specialists into competing architects of the entire AI data center infrastructure. Nvidia is expanding beyond GPUs into CPUs, software (via Hugging Face acquisition), and broad hyperscaler partnerships, aiming for a reusable AI factory model. Conversely, Broadcom focuses on custom silicon (XPUs) tailored for specific giants like Google and OpenAI, complemented by growing networking and VMware software revenues. Both companies benefit from record hyperscaler spending, with Nvidia targeting massive general-purpose scale and Broadcom offering purpose-built efficiency, making investment choices complex as both capture significant portions of the AI stack.
The artificial intelligence (AI) hardware race used to look like a two-lane highway: Nvidia (NVDA -0.03%) sold general-purpose graphics processing units (GPUs), while Broadcom (AVGO +0.32%) sold custom silicon, known as XPUs.
That story is becoming stale. After Nvidia and Broadcom each reported earnings, it's becoming increasingly clear that both companies look less like chip specialists and more like competing architects for the entire data center.
The question is no longer which company makes the better accelerator. Rather, investors should assess who is positioned to capture more of the AI infrastructure stack as hyperscalers continue to spend record sums on both custom ASICs and general-purpose chips.
Image source: The Motley Fool.
AI infrastructure is no longer a story of GPUs vs. XPUs
Nvidia is no longer just a pure-play GPU designer. During the second quarter, the company's data center segment generated $89 billion in revenue, up 117% from a year earlier. The interesting part is what sat alongside the GPU sales.
Grace CPUs, launched in 2021, have generated more than $5 billion in sales over the last year. Moreover, the new Vera Rubin CPUs are expected to generate roughly $20 billion in sales by next quarter, with management forecasting revenue to more than double in fiscal 2028 as the platform ramps.
Meanwhile, Nvidia's software layer continues to expand through CUDA, Nemotron, Cosmos, and Isaac. The company's recent acquisition of Hugging Face for $12.9 billion is the clearest manifestation of this software shift, underscoring how Nvidia is becoming the place where generative models are shared, not just the silicon they run on.
NASDAQ: NVDA
Key Data Points
Broadcom sits on the other side of the chip equation. Custom accelerators are the primary growth engine, with XPUs shipments comprising 73% of the company's $16.7 billion in AI semiconductor sales last quarter.
However, networking solutions are supposed to grow just as fast in the upcoming quarter. Meanwhile, infrastructure software generated $8.8 billion in revenue -- up 29% year over year. This growth is respectable and shows that the VMware acquisition is proving a transformative complement to Broadcom's semiconductor business.
Custom chips for Alphabet's Google Cloud, OpenAI, Meta Platforms, and Anthropic sit on top of Ethernet switches, optics, and software that Broadcom provides. CEO Hock Tan gives a straightforward pitch: If you design chips around one customer's model requirements, you can make it cheaper and more power-efficient than a one-size-fits-all GPU.
Inside the rack, Nvidia and Broadcom occupy different floors. Nvidia wants to be the reusable AI factory that any cloud provider or enterprise can plug into seamlessly. On the other hand, Broadcom wants to become a purpose-built engine for a handful of giant customers that will reorder by the gigawatt.
The subtle theme here is that hyperscalers are buying both solutions, making the investment choice between Nvidia and Broadcom much tougher.
Breaking down how Nvidia and Broadcom work with AI hyperscalers
During the second-quarter earnings call, investors learned that Nvidia expanded its relationship with Amazon Web Services (AWS) to include another 2 million GPUs through fiscal 2029. In addition, the company is also helping fund a 12-gigawatt data center campus that OpenAI is building in Ohio alongside SoftBank Energy.
When you layer in that every major cloud provider is already taking orders for Vera Rubin and the company's book of non-hyperscaler customers grew 138% year over year to $40 billion, it's nearly impossible to think about the infrastructure supercycle without Nvidia touching multiple sockets.
Broadcom's version of this story is almost as large as Nvidia but far more concentrated. Anthropic is on a path to grow from 1 gigawatt of Ironwood this year to 5 gigawatts of TPU v8i in 2027, giving Broadcom a line of sight to 10 gigawatts by 2028. OpenAI is narrowly trailing its rival, with 1.3 gigawatts of Jalapeno inference chips deployed in 2027 and more than 5 gigawatts in 2028. Meanwhile, Google Cloud remains a multiyear design partner for its Tensor Processing Units (TPUs), and Meta has three generations of custom MTIA chips booked through 2027.
Nvidia vs. Broadcom: Which is the better stock to buy?
The near-term math between Nvidia and Broadcom is pretty blunt. Nvidia guided current-quarter revenue to $108 billion, with no China data center compute included in this figure. Meanwhile, fiscal 2028 revenue growth is pegged at about 70%, with CFO Colette Kress calling it a "supply-constrained outlook." Unconstrained demand calls for growth above 70%, but memory, wafers, power, and shells are the bottlenecks through the end of next fiscal year.
Broadcom guided the current quarter to land around $34.8 billion in revenue, with AI semiconductors growing 236% year over year to $21.7 billion. For full-year fiscal 2026, AI revenue was raised to $58 billion -- reflecting 186% growth from last year.
Then the real tell is what management forecasts over the next two years. In fiscal 2027, AI revenue is expected to be $115 billion, surging to $230 billion by fiscal 2028. To put it another way, Broadcom is expected to double the size of its AI business for two consecutive years.
While growth rates favor Broadcom's AI business, breadth and cash generation favor Nvidia. This means valuation is what breaks the tie when picking between the two stocks. Nvidia trades around 28 times on a trailing price-to-earnings (P/E) basis and roughly 23 times forward earnings. Broadcom sits closer to a 46 times trailing P/E and about 31 times forward P/E.
In other words, investors are paying less for a company that is still guiding to 70% growth, with a broader customer set and a software distribution deal that Broadcom does not have.
NVDA PE Ratio data by YCharts
While custom silicon will keep winning market share from the frontier labs that can afford it, Nvidia is selling the scarcer asset: a comprehensive platform that many models can run on, backed by a balance sheet that can help finance and secure power and infrastructure years in advance. While its valuation multiples may not reflect it, Nvidia is already selling this platform across hyperscalers, neoclouds, and sovereign enterprises at once. That is why I think Nvidia has more upside than Broadcom and is the better buy of the two.