NVIDIA: The AI Powerhouse Transforming Global Tech

Emily Carter

Welcome back to Global Equity Research, the official podcast from Softgate Capital Research—the research arm of Softgate Capital, where we bring institutional-grade insights to anyone interested in the world of finance and macroeconomics. I’m Emily Carter, joined by my co-host, David Mitchell. As always, we’re here to help you cut through the noise with clear, actionable research.

David Mitchell

And we’re excited to do a deep dive today, Emily. If you’re new, be sure to check out our website at research.softgatecapital.com. That’s where we make our full research reports available—same level as what the big asset managers pay for, but actually affordable if you’re not managing a billion-dollar fund.

Emily Carter

Exactly. And today we’re spotlighting a company that’s basically become infamous in global markets for how much it’s changed the landscape—NVIDIA. We’re calling this one “NVIDIA: The AI Powerhouse Transforming Global Tech.” Let’s dig in.

Emily Carter

Alright David, just to level set, how did NVIDIA go from plastic video game cards to being, like, this backbone of the modern AI world?”

David Mitchell

Yeah, so—you know, it’s wild. NVIDIA started back in 1993 as a classic Silicon Valley story, totally revolutionized PC gaming with their GPUs. But—and this always impresses me—they didn’t stop there. Over time, they saw this new market of what they called “accelerated computing.” They shifted from graphics... to basically becoming the fundamental hardware for just about any heavy-duty computing you can think of. AI, data centers, robotics, healthcare—it’s in just about everything now.

Emily Carter

Mhmm. And it’s not just the chips, right? I mean, you hear so much about their software stack. CUDA, in particular. I love this part: it’s like, they quietly built this sticky ecosystem. Folks, imagine a digital arpeggio here—NVIDIA’s CUDA platform is this developer environment. If you want to run advanced AI workloads, odds are, you’re coding for CUDA. And if you’re coding for CUDA? Changing to a different chip is basically a multi-year undertaking. They’ve managed to make themselves indispensable, not just because of the hardware, but the developer lock-in.

David Mitchell

Right. It’s not just about selling a better chip. It’s about providing the full stack: hardware, software, even integrated systems like their DGX AI supercomputers. And now they’ve rolled out their own Grace CPUs, so it’s not only GPUs anymore—they’re making full AI superchips for the data center, basically pushing into areas that were once Intel’s turf. Their partnerships—VMware, Snowflake, every big cloud—just reinforce how central NVIDIA's platform has become.

Emily Carter

And that R&D machine is... honestly, it’s in a league of its own. Over $7 billion in FY2025. They plow high-margin profits straight back into developing new architectures—like the Blackwell GPUs that are coming. It’s this virtuous cycle, right? They lead, they profit, and then they double down. That's how you go from gaming to running the show in AI.

Emily Carter

So, let’s turn the page to the financials, because the numbers here are honestly jaw-dropping. David, can you give us some highlights—especially what changed from, say, the pre-AI-boom years to this wild AI surge?

David Mitchell

Of course. Back in FY2021, you were looking at revenue around $17 billion—mostly gaming and the early uptake in data centers. By FY2025? $130.5 billion in revenue. That’s nearly an 8x jump—over just two years, thanks mostly to the AI boom. Data Center alone was $115 billion last year. That’s 88% of their revenue now, and it grew 142% year-on-year. It’s insane, honestly.

Emily Carter

Right, and the gross margins followed. Mid-70s percent gross margins are pretty much unheard of at this scale. What floored me was the operating leverage: when revenue spiked with AI, EPS more than doubled, even as they ramped up R&D and hiring. GAAP operating income soared to $81.5 billion in FY2025, which is... bonkers. And let’s not forget—they’re sitting on over $40 billion in cash, have a massive $98 billion authorized buyback, with $62 billion unused as of late 2025. So yeah, even after a $4.5 billion inventory charge tied to those new export rules for China in FY2026, the bottom line is still up, thanks to this tidal wave in demand for AI silicon.

David Mitchell

Absolutely. And what I find fascinating is, we talked in our U.S. macro episode about how much tech spending is holding up broader growth. This is where you see it, right here: unprecedented AI infrastructure spending, flowing straight to NVIDIA’s bottom line. And the market’s priced in perfection: $177 per share, with a $230 target. It’s the most valuable company in the world by market cap right now.

Emily Carter

Just for scale, Q4 FY2024 revenue was $22.1 billion—that’s a 265% jump year-on-year in a single quarter. So as an analyst, if you had to zoom in: what metrics actually show NVIDIA’s operating leverage the best?

David Mitchell

I’d point to operating margin—it jumped from 16% to 62% in two years. Also, watch operating cash flow: $52 billion last year. When revenue goes up, expenses don’t rise nearly as fast, so the extra dollars just pour into profit and buybacks.

Emily Carter

But let's talk where the revenue really comes from. So David, what are the engines driving their margins now? Is it just one business line on fire and the rest are afterthoughts, or is there more to it?

David Mitchell

No, it’s pretty nuanced. The Data Center business is the obvious growth juggernaut, like we just mentioned, but gaming is still important. It’s the cash cow—keeps their developer ecosystem healthy, brings in stable cash flow, and honestly acts as a testing sandbox for cutting-edge tech, like ray tracing, that might eventually show up in AI applications. In FY2025, gaming was about $11.4 billion, roughly 9% of total revenue—so, smaller, but still serious money.

Emily Carter

Their Professional Visualization business—we’re talking about stuff like workstation graphics boards for simulation and design—that’s more niche, about $1.9 billion this year. Still, it’s a strategic adjacency because a lot of that tech trickles into the AI side. And auto—NVIDIA’s AI cockpit and self-driving solutions, mainly—climbed to $1.7 billion, with a pipeline close to $14 billion... though that’s a slower-burn story.

David Mitchell

What’s key is the fabless model. NVIDIA doesn’t own factories; they rely mostly on TSMC to make their advanced chips. That keeps their balance sheet clean, though it does create risk—one hiccup at TSMC, and there’s a backlog fast. But being fabless also means they can shift capital faster to R&D and ecosystem investments. If you want a sound-bite version, picture a bustling data center rack—rows of NVIDIA’s GPUs humming away, powering everything from ChatGPT to self-driving cars.

Emily Carter

And I want to touch on why the GPUs are the default choice for large-scale model training. It’s not just the raw silicon; it’s the software, developer ecosystem, optimization, and support you basically can't get anywhere else. Once a big player signs on, it’s very sticky—just ask any cloud provider, or those working with CUDA or DGX pods.

David Mitchell

Right, and the cross-pollination between gaming, AI, and pro visualization—same core architectures, just tweaked for each market—lets them scale like nobody else right now.

Emily Carter

Alright, so by now it sounds like NVIDIA is unbeatable, but the reality’s never that simple. Let’s dig into the risks. Who are they up against, and what could derail this story?

David Mitchell

Competition is fierce and getting worse. AMD is their main GPU rival—but still lags in AI, even after some wins with their MI300. Intel, well, they’re dabbling with discrete GPUs and some AI efforts, like Habana, but they’re way behind in performance and adoption. The bigger “in the weeds” threat is the custom ASICs—like Google’s TPUs or Amazon’s AI chips. Those are mainly for internal use, but if they ever get good enough or cheap enough, they could eat away at NVIDIA’s wallet-share at hyperscalers.

Emily Carter

Plus there are those AI chip startups—Graphcore, Cerebras—all looking for an angle, but none have reached NVIDIA’s commercial scale. The real lock is, again, CUDA and the software ecosystem. It’s not trivial to break that, even if you have a faster chip on some benchmarks.

David Mitchell

Absolutely. But it’s not just the competition—there’s a supply chain risk. NVIDIA is totally reliant on TSMC’s latest technology in Taiwan. That’s geopolitical risk and capacity risk, both at once. We’ve also got to talk about regulatory headaches. The new U.S. export controls took out $4.5 billion in inventory for China overnight in FY2026, and that could ramp up if global tensions worsen. And, look, margins are great now, but if competition heats up or customers push back, margin compression is a real possibility.

Emily Carter

And there’s potential gaps as well. NVIDIA’s still a minor player in CPUs compared to Intel and AMD, which could hurt in some all-in-one solutions, though they’re pushing into that with Grace. All these things could threaten the runway everyone assumes they have.

David Mitchell

If you ask me—as an industry analyst—how real is the threat from custom ASICs or AMD? I’d say, in the short-run, CUDA and the software ecosystem are massive barriers. But if a cloud giant’s chip suddenly leapfrogs on price-performance? Or if TSMC has a production snag? It could shift the whole landscape fast. The sound you’re hearing, by the way, is a low underscoring tension—these regulatory and supply chain risks are not hypothetical anymore.

Emily Carter

Okay, so let’s bring it together for investors—valuation, the scenarios, and the headline takeaways. David, how should people actually frame NVIDIA at these sky-high multiples?

David Mitchell

So, $177 share price, $4.3 trillion market cap, and a $230 target. Trailing P/E is around 60x, forward P/E closer to 30x for FY2027, so it’s rich, but context matters—earnings are growing so fast, the multiples compress quickly. Compare that to AMD or Broadcom, and NVIDIA’s got a much higher net margin—more than 50%, which is just... wild for this scale. Our view is that the PEG ratio tells a better story. With a near 50% earnings growth rate, even a PEG of 1.3x is ‘cheap’ for this kind of dominance. Our DCF cross-check has it comfortably supporting these levels—and possibly upside to $250-300 per share if the runway persists.

Emily Carter

Let’s do a “quick numbers” blast: share price $177, target $230, market cap $4.3 trillion, trailing P/E ~60x, forward P/E ~30x, gross margin ~75%, operating cash flow $52 billion, and $62 billion left in the current buyback program.

David Mitchell

And when you model the bear and bull cases? To the downside: if AI demand slows, or U.S.-China tensions cut out a market, or if margins dip, the multiple can compress. Worst-case, that’s probably a $120-140 scenario. On the upside, continued AI adoption and new verticals—software, automotive, edge—could push it well past $230-250 in a couple years. Buybacks are a wild card; they’ve got the ammo to boost EPS even if growth moderates.

Emily Carter

So, if you had to pick one metric to watch next quarter, what would it be—and why?

David Mitchell

I’d say data center revenue as a share of total—and gross margin. If both are sticky or up, the thesis stays very much intact. If we see any cracks in those, or major slowdowns at the hyperscalers, then we have to rethink the trajectory.

Emily Carter

And watch for those “Blackwell” GPU launches—if the next-gen chips launch strong, or enterprise AI adoption broadens, that could be, well, another leg up for the story.

Emily Carter

To wrap it all up—NVIDIA is basically the platform leader for AI infrastructure today, with exceptional margins and a history of nailing its execution. But you can’t ignore the risks: concentrated supply chain, regulatory shocks, and a valuation that leaves no margin for error. The thesis is robust, but, as always, the story can shift quickly in this industry.

David Mitchell

Absolutely. For more, check out the episode description or head over to research.softgatecapital.com for our one-page NVIDIA fact sheet—revenue mix, margins, cash and buyback stats—a competitor slide and risk checklist. If you want a deeper dive, let us know: we’re considering follow-ups on either supply chain resilience or a tour of the CUDA and Omniverse software ecosystem. Just send us your questions.

Emily Carter

And a quick disclaimer—everything here is for informational purposes only. It’s not investment advice, or a solicitation to buy or sell securities. As always, do your own research and consult with a licensed advisor before making investment moves.

David Mitchell

Emily, always a pleasure talking tech giants with you.

Emily Carter

Likewise, David. Thanks to everyone for joining us for Global Equity Research. Until next time, stay curious, stay sharp, and be sure to send us your questions for future episodes. Goodbye!

David Mitchell

Take care, everyone. See you next time.