selloff market pricing ai exposure rotation hardware software

The selloff last week revealed something important about how the market prices different types of AI exposure — and why the rotation from hardware to software could be both inevitable and violent.
Hardware AI (Nvidia, Broadcom, AMD, Micron) is a capex cycle bet. You're essentially betting that hyperscalers continue spending $50–60B per quarter on AI infrastructure indefinitely. The thesis is simple and powerful when rates are low: borrow cheap, build fast, the ROI comes later. When rates spike — as they did Friday with the 10Y hitting 4.54% — the "ROI comes later" part becomes a problem. Higher discount rates make future cash flows worth less. Hardware AI stocks get hit hardest because their earnings are largely front-loaded (the spending is happening now) while the monetization is still uncertain.
Software AI (Apple, Salesforce, ServiceNow, Microsoft) is a monetization cycle bet. You're betting that AI features drive incremental revenue — higher prices, lower churn, more seats, new services. The multiple is lower (40-60x vs 80-90x for hardware), but so is the sensitivity to rates. More importantly, software AI doesn't require the customer to keep spending on infrastructure. Once the AI feature is in the product, the revenue recurs.
The irony: the market spent six months rotating INTO hardware AI because the ROI story was clearer. Now, with rates rising and hardware stocks repricing, the rotation is reversing. Software AI — which looked boring compared to Nvidia's vertical chart — suddenly looks like the safer bet.
Apple sits at the intersection. 2.2 billion devices is the distribution advantage no hyperscaler can match. If Siri 2.0 drives even a 5% increase in services revenue, that's $5B+ in high-margin, recurring revenue that doesn't require a data center to scale.
Today's WWDC is the first real test of whether the rotation from hardware AI to software AI is real — or just a wishful narrative.