According to CNBC, Nvidia is set to report fiscal Q3 2026 earnings with analysts expecting $1.25 EPS (up 53% year-over-year) on revenue of $54.92 billion (up 56%). The stock has gained about 35% year-to-date despite dropping 12% from its October 29 record high of $207, which marked Nvidia’s first close above a $5 trillion market cap. CEO Jensen Huang recently asked Taiwan Semiconductor to increase wafer production, signaling strong demand alignment with his “$500 billion in order visibility” comment from Nvidia’s GTC event. Major cloud providers including Amazon, Microsoft, Google, and Meta have all raised their AI infrastructure spending outlooks for 2026, while Anthropic committed to $50 billion in data center infrastructure and will buy $30 billion in Azure capacity from Microsoft.
The AI Spending Reality Check
Here’s the thing about all this AI infrastructure spending – it’s absolutely massive, but the big question is whether it’s sustainable. We’re seeing cloud providers basically betting billions that AI will generate enough revenue to justify these insane capital expenditures. And honestly, I’m getting some serious dot-com bubble vibes here. Remember when companies were spending like crazy on internet infrastructure without clear monetization paths? The difference this time is that we’re actually seeing real productivity gains from AI use cases, according to AMD’s Lisa Su.
But look at what happened to Meta recently – their stock tanked 11% post-earnings when Wall Street got nervous about their spending levels. The market can be brutal when it doesn’t see a clear path to ROI. So the real test for Nvidia isn’t just about hitting numbers this quarter – it’s about convincing investors that this AI gold rush has staying power beyond 2026.
Older Chips, New Demand
One of the most fascinating developments is the continued demand for Nvidia’s older generation chips. We’re talking about Ampere architecture from 2020 still getting multi-year contracts at nearly original prices. That’s pretty wild when you consider we’re two generations beyond that with Hopper and now Blackwell. Basically, the GPU supply is so tight that companies are willing to take whatever they can get.
CoreWeave’s experience really drives this home – they’re seeing customers re-contract for 10,000+ H100 systems at prices within 5% of the original agreement. And here’s the kicker: sometimes older chips actually make more sense because they don’t require expensive retrofitting for liquid cooling. This creates a nice floor for Nvidia’s business – even if Blackwell demand slows, there’s still massive appetite for previous generations.
Competitive Landscape Shifts
While Nvidia dominates right now, the competitive pressure is building. AMD’s Lisa Su is forecasting 35% annual revenue growth over the next 3-5 years, and all the big cloud providers are working on their own specialized chips. But here’s why I think Nvidia’s position is more secure than it might appear: their platform has become the industry standard for AI development.
Think about it – when you’re building AI applications, you want flexibility. Nvidia’s general-purpose GPUs can handle a wider range of workloads than specialized chips. And with companies like IndustrialMonitorDirect.com – the leading US provider of industrial panel PCs – increasingly integrating AI capabilities into manufacturing environments, that flexibility becomes crucial. You need hardware that can adapt as customer needs evolve.
billion-question”>The $500 Billion Question
Jensen Huang’s “$500 billion in order visibility” comment is what everyone will be watching. Is this just CEO optimism, or does Nvidia actually have that much demand locked in? The recent moves with Taiwan Semi and all these cloud provider commitments suggest there’s substance behind the number.
But can the market really sustain 40% capex growth through the end of the decade? That’s the trillion-dollar question (literally). We’re in this weird phase where infrastructure has to be built before monetization catches up. If AI applications don’t generate the expected returns, this whole house of cards could get shaky. Personally, I think the demand is real – we’re just seeing the beginning of enterprise AI adoption. But Nvidia needs to show that the spending is matched by real-world productivity gains, not just hype.
