Nvidia? Fine. Broadcom is Better, Honestly.

Right. So, everyone’s going on about Nvidia. And yes, it’s true, the AI thing is… happening. Spending is up. Apparently, chips are the new black. And Nvidia, with its 90% market share, is the obvious choice. It feels safe. Which, let’s be honest, is the main criteria after the crypto debacle. Units of Cryptocurrency Lost: 12. Hours Spent Watching Charts: 9. Number of Panicked Texts to Friends: 24. But honestly? I’ve been doing some digging, and I think there’s something… better. Something a bit more… sensible.

It’s all about the CUDA software, apparently. Nvidia seeded it into universities. Clever. A whole generation of developers trained on their platform. It’s like… a benevolent dictatorship of chip design. Which is fine, I suppose. But it doesn’t exactly scream “long-term dividend growth,” does it? It feels a bit… precarious. Like building a financial empire on a foundation of… code. Which is probably solid, but still. My nerves.

So, I’ve been looking at Broadcom. AVGO. It sounds… solid. And boring. In a good way. The best kind of way. It’s like the sensible shoes of the tech world. Which, after a few years of chasing shiny objects, is exactly what I need. It’s about networking and custom chips. Networking. That’s a grown-up word, isn’t it? It implies stability. And custom chips. Apparently, these are the future. Less flashy than GPUs, but apparently much more efficient. And, crucially, better for the bottom line.

A Networking and ASIC Leader

Broadcom’s Tomahawk Ethernet switches are the go-to for big data centers. Apparently. I mean, I don’t know anything about Ethernet switches, but it sounds… important. And they also do fiber optics and digital signal processors. It’s all very technical. And slightly terrifying. But the point is, they’re managing the data flow. The actual infrastructure. Which, let’s face it, is where the real money is made.

But the really interesting thing is the custom chip business. ASICs. Application-Specific Integrated Circuits. It sounds complicated. And it is. But basically, they build chips to order. For companies like Alphabet. Apparently, they helped create the Tensor Processing Units (TPUs). Which are, apparently, very successful. And Alphabet’s spending a lot of money on them. Which is good. For Broadcom. And for my portfolio, hopefully.

And it’s not just Alphabet. Anthropic has apparently placed a $21 billion TPU order. Twenty-one billion. I need to lie down. And OpenAI is committed to deploying 10 gigawatts of custom chips. Gigawatts. It’s not a word I use every day. Apparently, each gigawatt equates to about $35 billion in chip costs. Which is… a lot. Broadcom generated slightly under $64 billion in revenue last year. Do the math. It’s… promising. Very promising. Although, I’m slightly afraid to actually calculate it.

It’s all a bit… overwhelming, isn’t it? But the point is, Broadcom is positioned to benefit from this AI boom in a way that feels… sustainable. Less about hype, more about actual, solid infrastructure. And, crucially, potential dividend growth. Which, let’s be honest, is the only thing that really matters. Will become disciplined long-term investor: 1/10. Still researching cat videos: 9/10.

So, yes, Nvidia is fine. It’s a good company. But Broadcom? Broadcom feels… right. It feels like a sensible, grown-up investment. And after a few years of chasing rainbows, that’s exactly what I need. It’s not glamorous, but it’s solid. And sometimes, solid is enough. Sometimes, solid is everything.

2026-02-21 10:02

A Networking and ASIC Leader

Read More