Why I'm Still Bullish on Nvidia After Earnings
Nvidia just crushed expectations. Revenue growth accelerated to +73% this quarter, up from +63% last quarter and +56% the quarter before. The trajectory is clear: AI infrastructure spending isn't slowing down, it's entering a new expansion phase.
And yet the stock is down 4% today.
The guidance alone should tell you everything. Next quarter's revenue is expected to accelerate further to +77%, and that figure excludes any contribution from China. Margins are guided at 75%, confirming that the Blackwell ramp-up headaches are firmly behind them.
The inventory scare that wasn't
Last quarter, a number of analysts and finance commentators flagged the sharp rise in Nvidia's inventory as a red flag, a potential sign that production was outpacing real demand.
Had they read the balance sheet notes more carefully, they would have seen that most of that inventory growth consisted of work-in-progress materials and components being staged for the upcoming Rubin chip launch, expected in H2 2026.
This quarter confirmed those concerns were completely unfounded. Inventory grew just 8% sequentially, while revenue grew 20%, a clear signal of excess demand, not oversupply. The inventory-to-revenue ratio is already compressing again.
The real story: Agentic AI has hit an inflection point
When an analyst pressed Jensen Huang on how much further Big Tech capex could realistically grow, given that collective spend is already approaching $700 billion and is ultimately constrained by cash flow, his answer was striking:
"Agentic AI has reached an inflection point."
This is the insight worth sitting with. Until recently, Big Tech was investing billions in Nvidia GPUs to build and train AI models. That continues. But now they're also deploying Agentic AI, systems that don't just answer questions like a chatbot, but autonomously execute entire workflows: writing code, running tests, managing tasks end-to-end.
Agentic AI requires dramatically more compute. And here's the key shift: more compute spend is no longer purely a cost, it's a revenue generator. When AI agents are automating entire job functions, capex becomes an investment with measurable returns in productivity and cash flow. That makes further capex growth sustainable over the coming years, which in turn keeps Nvidia's revenue pipeline intact.
Huang framed it this way: the world currently spends $300-400 billion on software. That figure will grow exponentially as AI requires vast compute capacity in exchange for automating entire categories of work. The progression runs from AI chatbots (where we started), to Agentic AI (where we are now), to AI embedded in robotics and physical manufacturing (what's next). In the meantime revenue accelerate again.
The energy efficiency argument
One legitimate concern with Agentic AI is energy consumption. Agentic workloads are significantly more power-hungry than training runs, and Nvidia's GPUs consume more power than purpose-built alternatives like Alphabet-Broadcom's TPUs or simpler GPU architectures from AMD.
Huang's response: the competition isn't purely a hardware race. Nvidia's advantage lies in the full stack. CUDA, TensorRT, and NVLink interconnects give its chips superior software efficiency, meaning customers generate more tokens per watt than on competing platforms. That's why, even on energy costs, Nvidia remains the preferred choice.
The thesis remains intact.
Comments
Post a Comment