8 min read
8 min read

Nvidia’s GPUs have become the gold standard in AI computing. If you’re training or running massive AI models, chances are Nvidia’s chips are powering the process.
But now, the AI industry is shifting. Hyperscalers think Amazon, Google, and Microsoft are realizing they don’t always need the flexibility of GPUs.
Instead, they’re exploring custom accelerators designed for specific workloads. That shift opens the door for players like Broadcom, which specializes in tailored chip designs and networking infrastructure.

Broadcom’s custom processors, known as XPUs, are positioned as strong alternatives to GPUs. Instead of one-size-fits-all computing, XPUs are built with hyperscalers for highly specialized AI tasks.
That customization can make them more efficient and cost-effective. With growing demand for AI-optimized hardware, Broadcom is stepping into a space where Nvidia has long dominated.
The big question: can Broadcom turn its XPU projects into real market traction and revenue growth to rival Nvidia’s staggering lead?

Tech giants that operate massive data centers are realizing that not every workload needs the brute force of GPUs.
Custom accelerators can outperform GPUs while using less power for routine or well-defined AI tasks. This is where Broadcom shines, designing chips optimized for specific use cases.
These AI hyperscalers spend billions annually on compute and are eager for alternatives that reduce costs. Broadcom’s partnership-driven approach could give it a strong foothold in these mega-scale data centers.
Though it doesn’t have Nvidia’s household recognition, Broadcom is no small challenger. With a market capitalization of around $1.4 trillion, it’s one of the seven largest U.S.-listed companies.
Its business spans semiconductors, cybersecurity, networking, and enterprise software through its VMware acquisition.
Broadcom isn’t starting from scratch; it’s leveraging scale, partnerships, and decades of chip expertise. Still, Nvidia is about three times larger in market value, showing its dominance in the AI hardware race.

Broadcom isn’t just building accelerators; it’s also a leader in connectivity. Its networking switches connect GPUs and XPUs across massive data centers, letting AI workloads be split, processed, and recombined efficiently.
Without these switches, AI clusters would be bottlenecked and less effective. Broadcom’s strength in networking could prove just as crucial as its custom chips.
By controlling the “glue” that ties computing clusters together, Broadcom positions itself as a critical player in the AI data center ecosystem.

Broadcom estimates its custom XPU market could reach between $60 billion and $90 billion by 2027. That’s from just three major customers today, with at least four more developing designs alongside Broadcom.
To put it in perspective, Broadcom’s trailing 12-month revenue was just under $60 billion. If these projections hold, XPUs could rival the size of Broadcom’s core business.
For investors, that signals a huge potential growth driver but only if Broadcom can deliver at scale.

Broadcom generated $4.1 billion in AI-related revenue in the first quarter alone. That may be small compared to Nvidia’s recent $26 billion quarterly revenue, but it’s a start.
More importantly, it shows Broadcom’s AI business isn’t hypothetical; it’s real and growing. Broadcom’s share of the pie could rise sharply as hyperscalers roll out larger deployments.
If Broadcom continues this trajectory, it may transition from being a diversified tech supplier to being seen as a core AI chipmaker.

The market has noticed Broadcom’s AI potential. Its stock trades more than 45 times forward earnings, an even higher multiple than Nvidia’s 42.
That premium means investors are already pricing in massive growth. But high expectations cut both ways. If Broadcom delivers, it could cement itself as a top AI play.
If it stumbles, the stock may face pressure. For now, investors are betting heavily that Broadcom’s AI accelerators will live up to the hype.

Nvidia remains the benchmark. In its latest quarter, Nvidia’s revenue grew 69%, compared to Broadcom’s 20% growth.
That kind of performance explains why Nvidia’s dominance feels unshakable. Broadcom may have a compelling growth story, but it is still far behind in execution speed.
To justify its premium valuation, Broadcom must accelerate revenue growth significantly. Otherwise, investors may question why it trades more expensively than Nvidia, the company it’s trying to compete with here.

Broadcom cites three major hyperscaler clients, including the likes of Google and others, for its $60–$90 billion AI opportunity, with up to four more clients reportedly designing XPUs with the company.
If Broadcom secures long-term, large-scale contracts with these players, it could lock in revenue streams for years. For investors, those relationships are the ultimate deciding factor in growth.

AI workloads require staggering amounts of compute, and GPUs alone aren’t always enough. Broadcom’s XPUs are designed to offload specific tasks and improve efficiency, helping hyperscalers manage scaling challenges.
Broadcom positions XPUs as essential co-pilots in data centers by complementing GPUs rather than replacing them entirely.
It’s a smart strategy: instead of trying to beat Nvidia at its own game, Broadcom offers tools that help AI giants stretch resources further, something every hyperscaler cares deeply about.

One of Broadcom’s standout innovations is the Jericho4 Ethernet fabric router. Built on a 3nm process, it can interconnect over one million XPUs across up to 60 miles.
That’s transformative for distributed AI infrastructure. Imagine AI clusters across cities functioning as one unified brain; that’s the vision.
Jericho4 strengthens Broadcom’s networking dominance and highlights why its role in AI extends beyond accelerators. It’s building the processors and the highways that keep AI traffic flowing.

Broadcom’s acquisition of VMware wasn’t just about software but also integration. By combining AI accelerators, networking hardware, and virtualization tools, Broadcom can offer hyperscalers a more complete ecosystem.
VMware’s expertise in managing workloads and virtual desktops complements Broadcom’s hardware push. For customers, that means smoother deployment of AI systems across complex infrastructures.
While Nvidia leans on CUDA and software ecosystems, Broadcom is crafting its integrated stack that blends chips, connectivity, and enterprise-grade software.

Despite its growth story, Broadcom’s stock valuation is a sticking point. Trading at over 45x forward earnings, it’s priced like a hypergrowth company, but its revenue growth trails Nvidia.
For some investors, this mismatch is concerning. Broadcom must consistently deliver AI-driven revenue acceleration to justify the valuation in the coming quarters.
Otherwise, enthusiasm could fade. As I see it, Broadcom’s biggest hurdle isn’t its technology; it’s proving to Wall Street that its AI opportunity is real, sustainable, and scaling.
Even with all the excitement around Broadcom, Nvidia’s entrenched position in AI is difficult to overstate. Its GPUs are the default for machine learning, and its CUDA software ecosystem locks in developers.
This combination creates a moat that competitors struggle to cross. Broadcom may win market share with custom chips for hyperscalers, but Nvidia still has unmatched dominance.
For Broadcom to truly rival Nvidia, it must scale its XPU sales and build a broader ecosystem advantage.
Take a look at how Intel is stepping up in the AI race with a massive dual GPU card packing 48GB of RAM.

Broadcom’s XPU push is bold, timely, and promising. With hyperscaler partnerships, networking leadership, and a clear AI strategy, it’s carving out a significant role in the AI era.
But calling it the “next Nvidia” is premature. Nvidia remains the undisputed leader, with unmatched growth and ecosystem lock-in.
Broadcom has the pieces to grow massively, but it must prove it can scale, deliver, and justify its sky-high valuation. For now, it’s a rising contender, not yet a champion.
Find out how TSMC and Broadcom are teaming up in ways that could threaten Intel’s long-standing dominance.
What do you think about Broadcom competing against Nvidia in making AI chips and GPUs? Please share your thoughts and drop a comment.
Read More From This Brand:
Don’t forget to follow us for more exclusive content on MSN.
This slideshow was made with AI assistance and human editing.
This content is exclusive for our subscribers.
Get instant FREE access to ALL of our articles.
Dan Mitchell has been in the computer industry for more than 25 years, getting started with computers at age 7 on an Apple II.
We appreciate you taking the time to share your feedback about this page with us.
Whether it's praise for something good, or ideas to improve something that
isn't quite right, we're excited to hear from you.
Stay up to date on all the latest tech, computing and smarter living. 100% FREE
Unsubscribe at any time. We hate spam too, don't worry.

Lucky you! This thread is empty,
which means you've got dibs on the first comment.
Go for it!