Just caught something interesting from Jensen Huang that's worth paying attention to. While the market's been hammering software stocks lately over AI fears, Nvidia's CEO is basically saying everyone's got this wrong.

Here's the thing - there's been this narrative going around that agentic AI is going to completely disrupt the software industry by replacing existing tools. Companies would supposedly just have AI agents build new software from scratch instead of paying for SaaS. But Huang thinks that's fundamentally misunderstanding how these systems will actually work.

His take is different. He reckons AI agents are going to primarily function as users of existing software tools, not replacements for them. Which means demand for those tools probably goes up, not down. The agents become power users operating on behalf of people, leveraging what's already out there. And software companies themselves can use AI to enhance their products further.

I think this is worth considering because Huang's got a unique vantage point here. Nvidia's GPUs power basically all the advanced AI training and inference happening right now, so he's watching this technology evolve in real time. When he says the market's gotten it wrong on software, that's not just speculation - it's informed observation from someone deep in the ecosystem.

The inflection point he's describing for agentic AI isn't about disruption through replacement. It's about acceleration through integration. That's a meaningful distinction that changes how you should think about software valuations going forward.

Software stocks have taken a brutal hit this year - we're talking about a $1.6 trillion market cap decline across the sector. Names like Salesforce, Workday, ServiceNow, Adobe have all gotten crushed. If Huang's reading is accurate, those kinds of valuations could look pretty attractive to investors with conviction in the actual thesis rather than the fear narrative.

Obviously there's still real risk that some businesses will face disruption. But the blanket bearish case on software seems to be missing Huang's point about how agentic AI will actually layer on top of existing infrastructure rather than replace it entirely. Worth thinking about if you're looking at this sector.