Perplexity CEO: AI spending will become more efficient very soon
Speakers
Arvind Srinivas
— Co-founder and CEO, Perplexity
Summary
Perplexity CEO Aravind Srinivas discusses AI spending efficiency, hybrid compute orchestration, and reaffirms NVIDIA's importance. He expects more efficient token spend and a future combining local and cloud AI inference. He dismisses the idea of selling NVIDIA and highlights small business AI adoption potential.
- AI spending is not slowing but will become more efficient.
- Inference currently runs on cloud servers; hybrid local/cloud orchestration is the future.
- NVIDIA remains essential for server-side AI compute and should not be sold.
- Perplexity is developing products to orchestrate models efficiently.
- Small businesses will adopt AI more resourcefully than large enterprises.
- AI enables building large companies with very few people.
- Regulatory and copyright challenges are facing the AI industry.
Trade Ideas
NVIDIA remains essential for AI compute.
NVIDIA will continue to do great because server-side compute remains essential for frontier intelligence. The shift to hybrid orchestration (local + cloud) does not eliminate the need for NVIDIA GPUs in data centers. AI inference will still run on servers alongside local compute, and NVIDIA's position is strong.
Up Next
This CNBC video, published May 29, 2026, features Arvind Srinivas discussing NVDA. 1 trade idea extracted by AI with direction and confidence scoring.
Speakers: Arvind Srinivas · Tickers: NVDA