I think the next AI leg may rotate from semis into enterprise software.

Trimmed $AMD $NVDA $INTC, rotated into $NOW, $CRM and speculative $PATH
$NOW cleanest agent infrastructure play, owns the workflow layer where enterprise AI gets orchestrated
$CRM levered to monetizing customer facing AI agents across sales, service and revenue workflows
$PATH higher risk asymmetric bet on automation evolving into agentic execution
Thesis: the next big AI spend may be enterprise deployment, not just model compute.
- AI is moving from chat interfaces to agents that execute work
- Enterprise adoption is shifting from experimentation to workflow automation
- The bottleneck is no longer model capability, it’s integration and change management
- Legacy systems make agent deployment hard, and that creates opportunity for software incumbents
- Compute and token budgets may become a new enterprise resource constraint
- The highest value use cases are moving from cost cutting to revenue generation
- Demand should rise for interoperable, agent agnostic software layers
- The future likely looks multi agent, not single platform
- Ironically, AI is increasing operational complexity before it reduces it
- Engineers increasingly move from coding software to operating agent systems
The market may be shifting from asking: Who enables AI training?
To asking: Who monetizes AI deployment?
$MSFT and $ORCL also on my radar.