←
Cynthia H1,530 posts
Cynthia H avatar

Cynthia H

@cyn

Comms @ Google.

◷ Joined November 2008▧ 51 Photos and videos
Amin Vahdat avatar
⟳ Cynthia H retweeted
Amin Vahdat@AminVahdat·Nov 9

Thrilled to share a big milestone from our AI & Infrastructure team at @Google. Today we’re expanding our 8th generation TPU portfolio with: • TPU 8i Max: Built for massive context windows and the highest compute density per rack • TPU 8tx: Purpose-built for secure, confidential inference This marks the first time we’ve introduced four distinct TPU systems in a single generation, in one year. For developers, hardware is only as good as the software stack. And we want you to be able to build with TPUs as easily and efficiently as we do at Google. You can now: ↳ Run native PyTorch & Hugging Face models via TorchTPU (zero code rewrites) ↳ Cut KV transfer latency >90% with open-source TPU Sync ↳ Eliminate idle compute waste with Agent Compute Engine Read more about our newest TPUs on our blog: bit.ly/3VfETPR I look forward to connecting with builders, researchers, engineers and organizations pushing the boundaries of AI and computing!

Cynthia H avatar
Cynthia H@cyn·Oct 8

Why I came back to Google: to work with some of the smartest people on the planet tackling some of the hardest problems possible, while maintaining humility the whole way.

One of our most beloved TD episodes (check out the YouTube comments). Amin is a true scientist ("genius level engineer") AND business mind. Inspiring conversation.

Post attachment
Sonya Huang 🐥 avatar
⟳ Cynthia H retweeted
Sonya Huang 🐥@sonyatweetybird·Oct 6

AI is driving the biggest infrastructure build-out in history, with @Google alone forecasting $200B of capex in 2026E. What are the real constraints on a system at this scale? Amin Vahdat is the chief in charge of Google's buildout. As SVP AI Infrastructure (AI^2), he role spans @GoogleDeepMind, @googlecloud, and the TPU/accelerator hardware roadmap, planning chips 2-5 years out and overseeing a data center buildout of epic proportions. Takeaways: — Why "goodput" per watt is the north star metric and designing around failures: At the 100,000-accelerator scale, something fails multiple times an hour, capacity has to double every six months, and as much of that comes from software and model optimization as from silicon. Hardware is a multiplier that lifts everyone above it — Why Google split TPU into two chips (8I for inference, 8T for training), and the calculus for when a workload is big and durable enough to specialize for — The "bitter lesson of chips," and why the TPU architecture hasn't fundamentally changed since v1 — How DeepMind and the hardware team co-design: delaying a tape-out by two weeks for the right model optimization, and using Gemini to design chips for future Geminis — Long-horizon agents as a new workload shape: no human in the loop to rate-limit requests, and CPU, networking and storage demand going "through the roof" alongside accelerators — Optical circuit switching: mirrors redirecting light to swap a failed TPU rack in milliseconds, without touching a fibe — Why Google prefers grid-connected over behind-the-meter power — Why 7- and 8-year-old TPUs are still at 100% utilization — Open standards, and why IP won the internet as the narrow waist of the hourglass — Orbital data centers: 1.4x the solar power, near-100% sunlight in sun-synchronous orbit, free-space optics between satellites, and "no fundamental showstoppers" — The 2036 rack: multiple megawatts, wheeled in, plug in water, power and fiber Timestamps: 0:00 – Introduction 1:47 – What makes a data center an AI data center 5:30 – Goodput, not FLOPS: holding yourself accountable when something fails every hour 11:52 – Doubling token capacity every six months, and where the gains actually come from 15:32 – The TPU bet: from a contrarian call in 2013 to splitting 8i and 8t 23:30 – The case for and against co-design 26:11 – Shoulder to shoulder with DeepMind: intercepting chips mid-flight 34:16 – Long-horizon agents change the shape of the data center 37:50 – Optical circuit switching and the state of networking 42:35 – Power is the binding constraint: utilities, gigawatts, and sizing a data center 49:23 – Training vs. serving clusters, seven-year-old TPUs, and open standards 58:29 – Orbital data centers and the supercomputer of 2036

1:04:26
Jason Ma avatar
⟳ Cynthia H retweeted
Jason Ma@JasonMa2020·Oct 5

When the New York Post calls your robot “the world’s most boring robot”, you know you are doing something right. 😭😂 This is actually the highest praise ever: when robots actually work, they are supposed to be boring, just calmly and reliably getting the job done.

Post attachment
Cynthia H avatar
Cynthia H@cyn·30 Sep 2025

Huge congrats to the @mai_agents team on their launch today! DTC brands just got a huge boost to growth arsenal, with AI agents to optimize performance marketing 24x7.

Yuchen Wu@yuchen__wu30 Sep 2025

Today we’re announcing @mai_agents — the AI platform that automates and optimizes performance marketing, and our $25M seed round led by @kleinerperkins. We’re launching now after nine months of building in stealth. We’ve already grown sales for companies like @NutritionFaktor by 40%:

3:41
Jason Ma avatar
⟳ Cynthia H retweeted
Jason Ma@JasonMa2020·29 Apr 2025

Introducing Dynamism v1 (DYNA-1) by @DynaRobotics – the first robot foundation model built for round-the-clock, high-throughput dexterous autonomy. Here is a time-lapse video of our model autonomously folding 850+ napkins in a span of 24 hours with • 99.4% success rate — zero human intervention • 60% human throughput speed • 4.3/5 quality ratings (set by the client) A thread on our motivation, insights and results:

2:34
Sharon Goldman avatar
⟳ Cynthia H retweeted
Sharon Goldman@sharongoldman·25 Mar 2025

Exclusive: @Instacart bought Lindon Gao's self-checkout startup for $350M. Now he’s teaming with a Google DeepMind alum, @jasonma2020 to build low-cost robots fortune.com/2025/03/25/exclu…

Justin Fiaschetti avatar
⟳ Cynthia H retweeted
Justin Fiaschetti@jcfiaschetti·14 Jan 2025

Ray is launching to space! This is a massive milestone for the whole @InversionSpace team as we make delivery from space a reality. I'd like to share some details about our company's first mission, launching NET Today, Jan 14 2025.

Post attachmentPost attachmentPost attachment
Marc Benioff avatar
⟳ Cynthia H retweeted
Marc Benioff@Benioff·3 Dec 2024

Q3 Results Salesforce Growth: FY25 $38.0B (guidance) FY24 $34.9B FY23 $31.4B FY22 $26.5B FY21 $21.3B FY20 $17.1B FY19 $13.3B FY18 $10.5B FY17 $8.4B FY16 $6.7B FY15 $5.4B FY14 $4.1B Salesforce Margin: FY25 32.9% (guidance) FY24 30.5% FY23 22.5% FY22 18.7% FY21 17.7% FY20 16.8% Salesforce Operating Cash Flow: FY25 $12.9B (guidance) FY24 $10.2B FY23 $7.1B FY22 $6.0B FY21 $4.8B FY20 $4.3B FY19 $3.4B FY18 $2.7B FY17 $2.2B FY16 $1.7B FY15 $1.2B FY14 $0.9B investor.salesforce.com/fina…

2:25
Austen Allred avatar
⟳ Cynthia H retweeted
Austen Allred@Austen·28 Jun 2023

Showed my wife ChatGPT this morning (she’s usually a late adopter). A couple hours later and she has it calculating nutritional intake for the family, then turning that to meal plans -> grocery list -> Instacart order. In <5 mins. “This used to hours every week. I might cry.”