CNBC Television
Sierra CEO Brett Taylor on implementing agentic AI: We're making agents that have memory
11/6/2025, 2:38:56 PM
Economic Summary
- Compute is the current rate-limiter for AI growth: both model training and inference consume substantial GPU resources, constraining deployment and scaling of AI services (relevant to NVDA, AMZN).
- AI agents can dramatically reduce per-interaction costs (phone calls ~ $25 today) and enable personalized engagement at scale, boosting conversions for businesses like Rocket Mortgage (RKT) and Madison Reed.
- Infrastructure investments are large, phased, and uncertain — depreciation cycles for GPUs and the timing when supply meets demand are key economic risks tied to long-term returns for vendors and cloud providers.
Bullish
- AI agents can personalize at scale, reducing per-interaction costs and improving outcomes.
- Companies using agents see materially higher conversion and engagement (example: Rocket Mortgage 4x).
- Agents with long-term memory can replace and outperform traditional customer service channels.
- Demand for compute remains the key bottleneck, supporting GPU and cloud infrastructure providers.
Bearish
- High compute and GPU costs could outpace revenue, straining margins for AI service providers.
- Uncertain depreciation cycles for GPUs risk large write-downs and stranded capital.
- Heavy infrastructure spending risks a bubble with some investments not yielding returns.
- Privacy, regulatory, and quality control issues could slow adoption in sensitive industries like healthcare.
Bullish tickers
NVDAAMZNRKTCIWWSIERRAOPENAIMADISON REED
Bearish tickers
NVDAAMZNSIERRAOPENAI
NVDA
Bullish
Primary beneficiary of surging demand for GPUs as compute is the bottleneck for AI training and inference.
Bearish
Risk of slower replacement cycles and potential write-downs if GPU depreciation is faster than expected; large-capital spending could be exposed.
AMZN
Bullish
AWS stands to gain from enterprises building AI infrastructure and deploying large-scale inference workloads.
Bearish
Massive infrastructure buildouts may include some inefficient investments; competitive cloud dynamics could pressure returns.
RKT
Bullish
Rocket Mortgage's Rocket Assist shows higher completion rates (4x), indicating strong conversion benefits from AI agents.
Bearish
Dependence on AI-driven funnels could backfire if agents underperform or face user trust issues.
CI
Bullish
AI agents can personalize healthcare interactions, potentially improving customer experience and operational efficiency.
Bearish
Healthcare deployments may face stricter privacy and regulatory scrutiny limiting rapid rollout.
WW
Bullish
Weight Watchers' AI agent showed higher empathy scores than call center coaches, improving customer engagement.
Bearish
Overreliance on AI coaching could erode human-led trust or effectiveness for some users.
SIERRA
Bullish
Agent Data Platform offers long-term memory and personalization, positioning Sierra as a platform for brand concierges.
Bearish
Startup faces competition, heavy infrastructure costs, and unproven long-term monetization at scale.
OPENAI
Bullish
Significant transformative impact across industries with demand limited primarily by available compute capacity.
Bearish
Rapid growth requires massive compute; high spending cadence could create financial stress or misallocated investments.
MADISON REED
Bullish
AI agents doubled guest bookings, showing clear short-term conversion benefits.
Bearish
Customer experience risks if AI personalization misfires or damages brand trust.
BUY.COM
Bullish
Mentioned as an example of how investment cycles can include both winners and losers.
Bearish
Used as analogy for overinvestment cycles; represents risk of ineffective players in investment cycles.
People mentioned
Brett TaylorWill Gadara