▶ 8:20 · $NVDA · Nvidia's NVL72 and Vera Rubin set premium benchmark but Grace CPU lacks high-core-count SKUs for agentic AI

episode briefing

US AI Policy Debate Reignited by Kimi K3, Where Anthropic is Lagging, AMD’s New Helios AI System

2026-07-22 · 6 company · 7 thematic

sentiment

2 bull1 bear3 neu

speakers

austin lions

Senior analyst at Creative Strategies covering semiconductor and AI infrastructure markets; provides competitive analysis on AMD, Nvidia, and data-center architecture.

corey weinberg

Corey Weinberg is a senior reporter at The Information who authored a profile of Anthropic CEO Dario Amodei and covers AI company business models.

laura bratton

Writes the Applied AI newsletter for The Information, reporting on enterprise AI adoption, coding agents, and frontier model competition.

quote

“Nvidia's had NVL72 um with the the Vera Rubin... Nvidia always has premium pricing... Nvidia just has one CPU and that CPU was always built in purpose of serving the GPU... they don't for example have very high core count um CPUs which in…”

—austin lions

now playing · $NVDA

$AMD···bullish· mediumaustin lions

AMD launches Helios rack-scale AI system to challenge Nvidia with CPU portfolio advantage for agentic workloads

AMD's first 72-GPU Helios rack system enables frontier model inference as a single integrated system, and its diverse EPYC CPU portfolio (vs Nvidia's single Grace CPU) positions i…

$NVDA···neutral· mediumaustin lions

Nvidia's NVL72 and Vera Rubin set premium benchmark but Grace CPU lacks high-core-count SKUs for agentic AI

Nvidia maintains premium pricing and rack-scale leadership with NVL72/Vera Rubin, but its single Grace CPU SKU optimized for GPU feeding lacks the high-core-count variants (e.g.,…

$MERCORneutral· mediumcorey weinberg

Mercor hits $2.8B annualized revenue run rate but 90% concentration in foundation model labs and 30% gross margins raise sustainability questions at $20B valuation

Mercor's human data labeling revenue accelerated to $610-620M gross in H1 2025 (30-35% net margin), yet 90% comes from a handful of frontier labs; investors betting on $20B valuat…

$ANTHROPICbearish· mediumlaura bratton

Anthropic lags OpenAI and Google in voice agents due to slower time-to-first-token; models optimized for coding not customer experience

Anthropic's frontier models are too large and slow for real-time voice/chat agents — NICE data shows smaller, cheaper models from OpenAI, Google, Qwen, and Mistral outperform on t…

$GOOGL···bullish· lowlaura bratton

Google's constant release of small models outperforms larger rivals in customer-experience voice agents

Google's strategy of frequently releasing smaller, optimized models (not just frontier-scale) gives it an edge in latency-sensitive customer-service voice agents where Anthropic's…

$TEAM···neutral· lowlaura bratton

Atlassian positions Jira as neutral AI coding intermediary to avoid enterprise lock-in to any single frontier model

Atlassian's new Jira AI capabilities let developers invoke GitHub Copilot, Claude, Cursor, and its own coding agent interchangeably, addressing CIO concerns about data sharing wit…

←