$CBRS · Google I/O 2026, Karpathy Joins Anthropic, and Cerebras’ $95B IPO | EP #256

episode briefing

extracted intelligence

Peter H. Diamandis

Google I/O 2026, Karpathy Joins Anthropic, and Cerebras’ $95B IPO | EP #256

2026-05-21 · 12 company · 10 thematic

sentiment

10 bull1 bear1 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.

dave blundin

Dave Blundin (DB2) is Peter Diamandis' longtime friend and MIT roommate, partner at Link Ventures, and described as a guru on exponential investing. He focuses on moonshot-stage companies in AI, robotics, and space.

alex wissner-gross

Computer scientist, physicist, entrepreneur and investor whose work spans artificial intelligence and complex systems. He is the founder and managing partner of Reified.

salim ismail

Salim Ismail is a serial entrepreneur and strategist who co-wrote Exponential Organizations (ExO 1.0 and 2.0). He leads the OpenExO community of 50,000 people in 150 countries and has worked with Fortune 500 companies (P&G, Siemens Energy, HP, Black & Decker) on edge innovation. He is launching the Organizational Singularity (ExO 3.0) as a living AI book via a Claude skill.

peter diamandis

Peter Diamandis hosts the Moonshots podcast with a panel of 'mates.' He founded XPRIZE and Singularity University, sits on boards of Trust & Will and Vesmark, and runs the Future Vision XPRIZE for positive AI futures. He frames discussions around the 'singularity economy' and exponential technologies.

quote

“We are somewhere between 15 and 20 times faster than the GPU on any inference problem... what you do is you divide the model very carefully... you pay a very small penalty for breaking it up into 4 8 16 20 chips. On the order of 2%.”

—andrew feldman

now playing · $CBRS

$CBRS···bullish· high· posandrew feldman

Cerebras wafer-scale engine delivers 15-20x faster inference via SRAM memory bandwidth breakthrough

Cerebras' dinner-plate-sized wafer-scale engine stuffed with SRAM overcomes the memory bandwidth wall, delivering 15-20x faster inference than GPUs with only 2% penalty for multi-…

$CBRS···bullish· high· posandrew feldman

Cerebras lands $20B+ OpenAI deal and AWS deployment after years of market indifference

After solving wafer-scale manufacturing (12 systems sold in gen1, 350 in gen2, thousands in gen3), Cerebras secured a $20B+ multi-year OpenAI deal and AWS data center deployment a…

$GOOGL···bullish· highdave blundin

TPU vertical integration from transistor to user experience is Google's unbeatable moat

Google's custom TPUs (2x power efficiency) combined with 1M+ TPU global clusters and 13 products with 1B+ users creates full-stack vertical integration no competitor can match.

$GOOGL···bullish· mediumalex wissner-gross

Google DeepMind only US frontier lab pursuing multimodality at scale

While OpenAI cut Sora and Anthropic focuses on code, Google DeepMind uniquely pursues omnimodal models spanning video, biology, and other modalities — a differentiated bet on moda…

$AMZN···neutral· mediumdave blundin

AWS dwarfs Amazon retail; Google competes on cloud not shopping vs Amazon

Amazon's retail business is now smaller than AWS; Google's universal cart attacks retail but the real battle is GCP vs AWS for AI compute infrastructure where both are investing h…

$META···bullish· mediumsalim ismail

Meta leads smart glasses race after Google abandoned Glass; Apple playing catch-up

Meta iterated for years spending billions on Ray-Ban smart glasses while Google abandoned Glass to enterprise then entirely — Meta now owns the consumer smart glasses lead with Go…

$ANTHROPICbullish· mediumandrew feldman

Anthropic wins Karpathy as talent magnet; focused on code generation enterprise strategy

Anthropic's recruitment of Andrej Karpathy (OpenAI co-founder, ex-Tesla FSD) signals its status as a top-tier frontier lab alongside OpenAI, with a differentiated focus on code ge…

$NVDA···bearish· mediumandrew feldman

Nvidia GPU architecture optimized for throughput not latency; $4M NVL72 serves 1-2 users at 200 tok/s

Nvidia's GPU architecture generates massive aggregate throughput but fails at per-user latency — an NVL72 at 35 tok/s is painfully slow, and at 200 tok/s serves only 1-2 users per…

$SPCX···bullish· mediumandrew feldman

Starship 100x cheaper launch enables orbital compute; Terafab 15-20 year project per Cerebras CEO

SpaceX's Starship could reduce launch costs 100x enabling orbital data centers, but building semiconductor fabs in space (or even on Earth via Terafab) is a 15-20 year endeavor du…

$TSM···bullish· highandrew feldman

TSMC remains undisputed manufacturing leader; Samsung and Intel years behind despite same ASML tools

TSMC's generational fab learning and yield expertise create an insurmountable moat — even with identical ASML equipment, Samsung and TSMC are not at the same node, and Intel has s…

$OPENAIbullish· mediumdave blundin

OpenAI raising $120B+ and exploring space compute partnerships (Blue Origin, SpaceX) for Dyson swarm

OpenAI's massive capital raise and Sam Altman's forward-looking strategy include securing heavy launch capacity (Starship, New Glenn) and potentially building orbital data centers…

$GOOGL···bullish· highpeter diamandis

Google's 6x CapEx to $180-190B validates AI-first vertical integration thesis

Google's massive CapEx increase (from $31B to ~$185B) combined with vertical integration from TPUs through 13 billion-user products creates an unmatched AI-native operating system…