▶ 187:00 · $FIGURE · Foundation Capital Steve: Humanoid robots wrong form factor; industrial automation wedges preferred

episode briefing

Cerebras IPO, Warshtime, VC Infighting, H100s in China, Robot Monkey Business

2026-05-14 · 13 company · 8 thematic

sentiment

8 bull2 bear3 neu

speakers

andrew feldman

Leads Cerebras, a public company (CBRS) building wafer-scale AI inference chips; previously co-founded SeaMicro (acquired by AMD); describes $25B backlog and global data center buildout at unprecedented scale.

amy reinhardt

ben hilac

eric vishria

Eric Vishria is a General Partner at Benchmark Capital, where he has led investments in companies like Cerebras, Fireworks AI, Sierra, Sunday Robotics, New Lantern, and Benchling. He focuses on AI infrastructure, applications, and hard tech. Previously co-founded and led Rockmelt (acquired by Yahoo).

steve

doug laughlin

quote

“I would say I I am generally um not a big believer in the humanoid uh approach. I think there are use cases uh for example in the home companionship. Yeah. Uh and and even in that case it's it's a bit of a stretch. Uh I think you need to t…”

—steve

now playing · $FIGURE

$NFLX···bullish· medium· posamy reinhardt

Netflix Ads President: Ad tier expanding to 15 countries, building proprietary stack, full-funnel focus

Netflix's ad-supported tier is scaling rapidly with proprietary tech stack launched a year ago, expanding to 15 new countries in 2025, targeting enterprise top 400 advertisers wit…

$RAINDROPbullish· medium· posben hilac

Raindrop launches open-source local agent observability tool for self-healing loops

Raindrop solves the missing local observability layer for agent development: developers cannot see traces locally before push, and coding agents lack access to execution traces fo…

$CBRS···neutral· highdoug laughlin

SemiAnalysis: Cerebras IPO succeeds but SRAM scaling wall limits large model inference

Cerebras wafer-scale chips deliver exceptional fast inference for sub-trillion parameter models, but SRAM scaling has stalled (WSE-3 only 10% memory increase over WSE-2), making i…

$NVDA···bullish· mediumdoug laughlin

SemiAnalysis: Nvidia Blackwell GB200 racks enable disaggregated inference with Groq SRAM offload

Nvidia's GB200 rack architecture allows passing activations to Groq LPU racks for extreme decode speedup, exemplifying the trend toward disaggregated inference where prefill (comp…

$GROQbullish· mediumdoug laughlin

SemiAnalysis: Groq LPU SRAM offload enables disaggregated decode acceleration in GB200 racks

Groq's LPU architecture with extreme SRAM bandwidth fits as a decode accelerator in disaggregated inference pipelines, receiving activations from Nvidia GB200 prefill; moving data…

$AMD···bullish· mediumdoug laughlin

SemiAnalysis: AMD MI300 rack scale-up executing, inference serving demand to follow Nvidia roadmap

AMD is focused on rack-scale execution for MI300/MI350; Lisa Su will figure out inference serving and likely match Nvidia's disaggregated roadmap with a fast SRAM offload FFN chip…

$INTC···neutral· highdoug laughlin

SemiAnalysis: Intel foundry stock ahead of technical turnaround; government deal a stroke of genius

Intel 18A process is good enough and customers (Apple, Musk) are committed due to US government top-down mandate, but historical execution risk remains; stock price has outpaced t…

$TSM···neutral· highdoug laughlin

SemiAnalysis: TSMC clean room bottleneck drives incremental not revolutionary capacity expansion

TSMC is capacity-constrained by clean room lead times (3-5 years); they expand methodically and incrementally, not revolutionarily, creating persistent shortages that benefit over…

$WIX···bearish· highunknown

Wix stock down 45% in 2025 as vibe coding AI website builders disrupt template model

Wix shares have collapsed from $250 to $52 as AI coding tools (vibe coding) enable one-shot website generation, destroying the template-based moat; Squarespace taken private at $7…

$CBRS···bullish· high· posandrew feldman

Cerebras CEO Andrew Feldman: Fast inference will become the entire market, like broadband replaced dial-up

Real-time AI interaction will drive all inference demand to fast models; slow inference has zero market size, analogous to dial-up vs broadband. Cerebras focuses on wafer-scale ch…

$CBRS···bullish· high· poseric vishria

Benchmark's Eric Vishria: Cerebras Series A thesis — GPUs suck for deep learning, infrastructure > applications

In 2016, Benchmark invested because Cerebras argued GPUs were merely 100x better than CPUs for deep learning, not optimal; the firm bet on infrastructure over unpredictable applic…

$SOL···bullish· medium· possteve

Foundation Capital: Solana investment driven by hard-systems thesis shared with Cerebras

Foundation Capital backed Solana because Anatoly Yakovenko (ex-Qualcomm, Dropbox) recognized the firm's appetite for hard distributed systems problems, evidenced by their Cerebras…

$FIGUREbearish· high· possteve

Foundation Capital Steve: Humanoid robots wrong form factor; industrial automation wedges preferred

Humanoid form factor is suboptimal for most industrial tasks (e.g., moving pallets); robotics should start with purpose-built automation (forklifts, washing machines) that diffuse…

←