Peeyush Ranjan

Worth paying attention to... Anthropic vs. Cursor. While Cursor was paying Anthropic hundreds of millions in API fees to power its AI code editor, Anthropic seems to have studied exactly what developers wanted, then launched Claude Code to "eat" their biggest distribution partner. The result. Claude Code hit $1B ARR in just six months and has already surged past $2.5B, proving that when you're a "wrapper" startup, you're often just paying your provider to learn how to replace you. Keep an eye on Cursor layoffs...

89

13 Comments

I spoke with Bay Area founders, engineers and product leaders about the rise of “vibe coding” and what it could mean for tech jobs in San Francisco and beyond. My latest for the San Francisco Business Times. https://lnkd.in/eHqkRs6H

22

It was a pleasure to chat with Rakesh Agrawal over at PCMag on his recent article that dug into how Waymo is changing late-night rides, especially for women. A few key takeaways: 🛡️ Many women prefer Waymo at night because there’s no driver involved, which reduces the risk of uncomfortable or unsafe interactions. 💸 That added sense of safety often comes at a premium. In one example, Uber and Lyft rides were under $20, while Waymo was over $50 for the same trip. 🚗 Waymo has a fixed fleet. Unlike Uber and Lyft, it can’t add more cars when demand spikes, so prices rise or rides disappear. 🎀 The result is an indirect “pink tax.” Men are more likely to choose cheaper human-driven rides, while women are more likely to pay extra for safety. ⚖️ Ride-hailing platforms are trying to close the gap with features like women-preferred driver matching, but smaller pools can also mean higher prices. ❓ The big question: should safety come at a premium, or should companies like Waymo rethink pricing so safer options aren’t effectively a luxury?

10

5 Comments

When asked whether "AI is taking up all the oxygen," I generally respond: "Yes. Yes it is." Here's a decent indicator as to why: https://lnkd.in/ga7MCAPa

14

100 founders, builders, researchers, investors and thinkers. One big question: how to solve some of the largest challenges building at the frontiers of AI. From physical compute scaling constraints to applying AI to unlock civilizational progress from material science to drug discovery. Camps, hikes and heated debates on human x AI co-existence: this was Climbing Hard AI Peaks. Thanks to everyone for coming out and making it so special. We learned a lot from all of you. Big shout-out to our founders working on some of the biggest AI unlocks, the many researchers and builders from Center for Digital Technology and Management (CDTM), ETH Zürich, Stanford University and special guests from Anthropic, OpenAI, Google DeepMind and more. David Byrd Michael Wax Jason D. Whitmire Chad Fowler Jonah Anders Kaplan Christine Fuchs Sam Harrison Mehdi Ghissassi Antoine Moyroud Wulfie Bain Eva Spannagl Nathan Gruber Judith Dada Juan Benet Zoe Weinberg Haya Hanna Sophia Kalanovska, PhD Clayton Mellina Ricardo Sequerra Amram Evan Phoenix Hendrik Dietz Katie Hewitt Molly Mackinlay Omer Shlomovits Zied Bahrouni

129

16 Comments

Vertical AI feels like a horserace 🏇 but after stepping back and thinking about the current state of market, I think what's equally unique is you have three different models competing at once. I think about it through this lens: 1️⃣ AI-enabled SaaS: the familiar path Take a proven vertical category and supercharge it with AI. Clear ROI, legible GTM, easy for VCs to understand. But real questions around differentiation and long term moats. 2️⃣ AI Roll-Ups: PE gets a facelift Acquire businesses, deploy AI to improve unit economics, and own the data + operations. Powerful if executed well, but operationally and financially complex. Big open questions around end state valuation. 3️⃣ AI-native industry disruptors: Become the thing Don’t sell software. Don’t buy incumbents. Build a fundamentally new version of the industry from scratch. Highest upside, but probably the highest casualty rate. You have to choose the model where the physics match the market. Each model has different capital needs, metrics, risk profiles, and founder skill requirements. I did a deep dive on each model 🤿 : ▪️where each model wins (and fails) ▪️how data dynamics actually differ across them ▪️why valuation expectations may diverge more than people expect ▪️and where I think the biggest outcomes will come from... As always, if you are building, investing, or just trying to make sense of vertical AI landscape, I always love to jam. Full post here: https://lnkd.in/g9a_9rtj

50

8 Comments

Martin Eberhard played a key role early at Tesla and even brought in Elon Musk. Then he was pushed out of the company for pennies. Founders get most of the attention in Silicon Valley, but the ecosystem is built on the people who take a chance and join early stage ventures. Unfortunately, those same people are often the easiest to cut out. Founders can dilute early employees, change vesting terms, and restructure equity with little effort and zero consequences (you can literally do it in Carta in 5 minutes). Daniel Kottke was a close friend of Steve Jobs and helped build the Apple I and II computers. He got no equity (Steve Wozniak eventually gave him a few of his own shares). Brian Acton co-founded WhatsApp and stayed through the Facebook acquisition. When he left Facebook over privacy concerns, he had to walk away from $850 million in unvested stock. There’s more awareness now around how vesting works, what happens during a liquidity event, and how to think about re-ups and dilution. That’s good. But the truth is, if someone really wants to screw you, they can. It still comes down to one thing: trust the people you work with. Or don’t work with them at all.

85

8 Comments

Greylock is leading the seed in SuperMe, the AI-native professional network. Most expertise is not yet legible to LLMs. Professionals demonstrate expertise in private contexts: product docs, meeting notes, and most often - in their own heads. SuperMe is building a way to extract and structure these insights into AI profiles, grounded in an expert's real words and personality. And it is network-first; the more conversations one's AI profile has, the more opportunities for experts to reinforce or create new insights -- expertise compounds. This is one of the strongest examples of founder-market fit we've seen at Greylock. Casey Winters is the most in-demand product & growth advisor that marketplace operators (self included) wish they could hire. And Ludo Antonov built & led engineering teams at some of the most iconic network businesses over the past decade: Whatnot, Lyft, and Pinterest (where he worked closely with Casey). The two started this company out of our offices, and have until now been quietly building a highly curated network with profiles like Elena, Lenny, Brian, etc. We see a unique opportunity to create the professional network graph in a world where likes & follower counts have lost meaning. More on our investment below:

286

13 Comments

Recently, a friend told me he was "long on compute." My first reaction — obviously, isn't everyone long on compute? Then I started actually using Claude more seriously. I don't think the market has priced in what personal agents do to compute demand. People are already buying dedicated Mac Minis just to keep their agents running 24/7. That's not a niche hobbyist thing. That's a signal. We're used to thinking about the compute build-out in terms of hyperscalers and GPU clusters. But the real story might be at the individual level — billions of people running personal agents in the background, continuously. Every individual consuming data center-scale compute. Not shared, not enterprise. Personal. We may be a lot earlier than we think.

24

10 Comments

10 Years In VC: What I Got Wrong at the Start "VCs are like LLMs - they feel compelled to give you an answer if you ask a question, even if they have no knowledge or experience of the topic." A founder recently said this to me, and after laughing out loud for real, it made me reflect on my own experience as a VC. After a decade in venture capital, I've learned a counterintuitive truth of this business: The best thing you can do for a founder is often nothing at all. When I started, I thought my job was to help. To have answers. Hustle like hell to find them if I didn’t have them. To share “wisdom” from other companies I’d seen. It came from a good place—wanting to add value. But it was wrong. Because most advice is context-poor. What worked in one company, in one market, in one decade…is often irrelevant, or even harmful, in another. Founders need real, relevant perspective—from people who are in the arena building right now, or customers adopting today, so they can decide for themselves. So now I spend my time helping by keeping my network more refreshed than theirs so I can get them to these people faster. Or I help by just listening when they need it. Another uncomfortable truth we don't always want to admit as VCs: exceptional founders don't need us. They will stop at nothing to achieve their mission. We might help them get a key hire or a customer faster, but they would have gotten there anyway. The reality is liberating once you accept it. Instead of trying to be indispensable, you can focus on being genuinely useful in the moments that matter. This shift has required growth on my side too. In a session with a coach, I once realized my own sense of self worth came disproportionately from feeling like I helped someone. I was optimizing for feeling useful, not founder success. The two aren't the same thing. I now think about it as finding exceptional people and enabling them to truly thrive. How you enable someone to thrive is so different and specific to that person. But largely, if it’s something I can uniquely help with, I’ll push like you wouldn’t believe - just ask my founders. If not, I leave teams alone to build and minimize annoyance, which they truly appreciate. The best investors are probably like great jazz musicians - they know when to play and when to stay quiet. My approach today: - Connect, don't correct. Prescribe almost never. Always connect founders to someone with recent, relevant experience. - Listen, don't lecture. The founder journey is isolating & hard. Sometimes the most valuable thing to offer is listening without judgement. - Enable, don't dictate. Create the conditions for amazing founders to thrive. And then… just get the fuck out of the way. Great venture capital isn’t about having all the answers. It’s about helping founders unlock their own. Founders — what’s the most useful thing a VC has ever done for you? VCs — what’s the most counterintuitive thing you've learned about venture?

179

62 Comments

Facebook is in talks to hire former GitHub CEO Nat Friedman and his VC partner Daniel Gross, could spend $1B+ to buy out their share in their VC fund. Gross is also a co-founder and supposedly the CEO in Ilya Sutskever's Safe Superintelligence

13

4 Comments

I spoke w/ Cloudflare co-founder & CEO Matthew Prince for the latest ep of the Slow Hunch podcast. Since 2010, Matthew and his team have built Cloudflare into one of the most important companies on the internet: powering and protecting vast portions of global traffic. I’ve known Matthew since Union Square Ventures’ investment in Cloudflare’s Series C back in 2013, and it was really fun tracing his slow hunch, right from those early days. We covered: - how he met co-founders Michelle Zatlyn and Lee Holloway - the early experiments and risks that shaped Cloudflare - stories from the company’s pre-IPO days - the decision to make encryption free - how their infra ended up running 2 of the internet’s 13 root servers - his thoughts on AI (esp the transition from a search-driven internet to an answer-driven one) Hope you enjoy!

278

6 Comments

GenAI Processors from Google DeepMind - an awesome elegant abstraction for building real time streaming asynchronous agents (and more) 🏆 A special shout-out to André Elisseeff - the main developer and tech lead, and top favorite friend and human 😘 Cool illustrative examples from their announcement (https://lnkd.in/e8KJhUrv): For example, you can easily build a "Live Agent" capable of processing audio and video streams in real-time using the Gemini Live API with just a few lines of code. In the following example, notice how input sources and processing steps are combined using the + operator, creating a clear data flow (full code on GitHub): from genai_processors.core import audio_io, live_model, video # Input processor: combines camera streams and audio streams input_processor = video.VideoIn() + audio_io.PyAudioIn(...) # Output processor: plays the audio parts. Handles interruptions and pauses # audio output when the user is speaking. play_output = audio_io.PyAudioOut(...) # Gemini Live API processor live_processor = live_model.LiveProcessor(...) # Compose the agent: mic+camera -> Gemini Live API -> play audio live_processor = live_model.LiveProcessor(...) live_agent = input_processor + live_processor + play_output async for part in live_agent(streams.endless_stream()): # Process the output parts (e.g., print transcription, model output, metadata) print(part) You can also build your own Live agent, leveraging a standard text-based LLM, using the bidirectional streaming capability of the GenAI Processor library and the Google Speech API (full code on GitHub): from genai_processors.core import genai_model, realtime, speech_to_text, text_to_speech # Input processor: gets input from audio in (mic) and transcribes into text input_processor = audio_io.PyAudioIn(...) + speech_to_text.SpeechToText(... ) play_output = audio_io.PyAudioOut(...) # Main model that will be used to generate the response. genai_processor = genai_model.GenaiModel(...), # TTS processor that will be used to convert the text response to audio. Note # the rate limit audio processor that will be used to stream back small audio # chunks to the client at the same rate as how they are played back. tts = text_to_speech.TextToSpeech(...) + rate_limit_audio.RateLimitAudio(...) # Creates an agent as: # mic -> speech to text -> text conversation -> text to speech -> play audio live_agent = ( input_processor + realtime.LiveModelProcessor(turn_processor=genai_processor + tts) + play_output ) async for part in live_agent(streams.endless_stream()): …

31

4 Comments

CRV Security: Request for Startups I never know if this actually works for our friends over at YC but figured we'd try. Here's what we want to fund in 2026! 1. Golden Artifacts: Think Chainguard but more broad. Artifact attestation exists for open source. Almost nothing exists for internal software — especially the vibe-coded tooling now running in production. We want the company building cryptographic proof of secure software delivered from secure artifacts: who built it, how, and whether it was reviewed. If more things are being yeeted into the world via Claude Code (myself included), this feels like an issue. 2. MCP & Agentic Security: Agents are getting real credentials and taking real actions. The security posture of most orgs around this is basically zero. That changes fast. You'd never give an employee hardcoded API keys or write access to your email without supervision/trust. Why give it to agents? 3. AI Governance: Boards are asking CISOs to account for AI risk. CISOs have no good answer other than "Palo has a module" 4. Next-Gen Endpoint: CrowdStrike was built for a world of static binaries and human operators. AI workloads, cloud-native infra, and AI-assisted attackers need a new architecture. The category is ready to be reinvented. 5. Networking in the AI Era: Zero trust was designed for humans. What does network security look like when the entity requesting access is an agent? Nobody's really solved this. 6. Email Security + Next-Gen Phishing: LLMs have made spear phishing infinitely scalable. I've never truly understood why Abnormal and KnowBe4 aren't one company. Maybe this time it's different. 7. Frontier Security Lab: We'd back a credible, well-staffed lab focused entirely on red-teaming models and setting the evidentiary standard the industry needs as LLM built apps become the norm. 8. Dependency Security: That Actually Remediates Malicious and vulnerable dependencies are a top attack vector. The tooling is mostly noise — scanners that don't close the loop. The winner here ships fixes, not just alerts. 9. Critical Infrastructure Cyber: Data centers, satellites, power grids, undersea cables. The physical backbone of the internet is increasingly exposed and wildly under-defended. We have data centers in space, for God's sake. Surely we need better cyber for critical infrastructure? 10. PAM for the Modern Era Legacy: PAM was built for static roles, human users, on-prem directories. Cyberark was founded in 1999.....Agents, ephemeral workloads, and cloud-native infra have broken all of those assumptions. Is anyone rebuilding this from scratch? If you're building in any of these areas — or something we haven't thought of — reach out. james@crv.com

451

49 Comments

Happy Monday :) SF AI-Native Operator Takeaway #1: Forward-deployed engineers and how AI companies are ~actually~ getting built. One of the biggest differences between AI-native companies and traditional SaaS right now isn’t just the technology or the model. It’s how and where the product actually gets built. The strongest AI teams I met aren’t optimizing pitch decks or even demo environments, they’re building inside customer environments. That’s why forward-deployed engineers and implementation strategists keep coming up. In practice this means: engineers sitting directly with customers; shipping integrations, workflows, and edge cases in real time; product managers working with those FDEs to understand what needs to be done; and learning what actually matters to the customer before anything gets productized. This flips the old SaaS playbook. Traditional SaaS assumed engineering was scarce. You build once, sell many times, and customize through Sales and CS. When bespoke engineering was required, it was often rational to walk away. AI-native teams are operating under a different assumption: engineering and GTM are treated as equally flexible. With that, services are often the wedge to becoming a platform, not something to run from. The vision is still to be a platform, but the momentum starts with high tough delivery. That said, there’s an important caveat that came up repeatedly. The forward-deployed model only really works when contract values can support it, and yet right now, it feels like almost everyone is trying to use it. Forward-deployed work drives speed, but it also blurs: ➖ Product versus services ➖ Pricing models, such as software plus implementation or usage plus services ➖ Gross margins once delivery, support, and compute costs normalize The best teams aren’t blind to this, but they still use the approach to rapidly build product. They focus less on feature validation and more on customer willingness to pay for something (before building it!), as well as what multiple customers repeatedly ask for versus what is truly bespoke. With that, they decide what should stay services versus what makes it into the core product. In this rapidly changing time, speed beats elegance, but only if teams are honest about what they’re actually selling and what the economics can support. Forward-deployed teams aren’t a scaling strategy on their own, but they’re an incredibly compelling learning strategy. And right now, learning (and implementing) faster than everyone else is the advantage. Next up: how PLG and GTM are changing in the AI landscape, and why many teams are fixing the wrong problem. #FDE #AI

14

2 Comments

"LLMs aren’t the only kind of AI", "One of the biggest, most important one is physical AI. AIs that understand the laws of nature" Not me, but Jensen Huang at CES26. The switch from being able to calculate language and simulate talking about things to understanding how things do things will unlock vast value. Matrix multiplication and reinforcement learning is transforming how knowledge work is done. The next step will impact how work is done at all. For the builders who figure it out and the investors who get it, LLMs will be the warm-up act.

36

2 Comments

AI agents without memory are stateless, inefficient, and limited. That becomes even more important as enterprises move from single-agent workflows to multi-agent systems, where context needs to persist across agents, sessions, tools, and LLM vendors. Excited to join David Joy from Cockroach Labs on May 19th for a deep dive on architecting persistent memory for multi-agent systems. 🚀 Timely topic for us: Memori Labs just released its new agent-native memory infrastructure, which automatically structures memory from agent trace and execution - not just natural language conversation. That means agents can remember what they did, not only what they said. We’ll also be sharing some exciting demos and perks for those who attend. Join us Tuesday, May 19th at 10AM PT. Register here: https://lnkd.in/gxWfxAxh

15

1 Comment

A new economic primitive is emerging in the software world, and it all comes down to one thing - tokens. Jensen Huang calls it the "Token Economy", and explains it plainly on the latest earnings call: "In this new world of AI, compute equals revenues. Without compute, there's no way to generate tokens. Without tokens, there's no way to grow revenues." The token is becoming to AI businesses what the unit of production is to manufacturing - a priceable output with a priceable cost structure. If this becomes the foundation of the new economy, then companies building on top of the models (i.e., App Layer) aren't just software companies anymore. They still rely on product quality to win, but under the hood they operate an arbitrage game of cost-per-token vs. revenue-per-token. In other words, token economics becomes an integral part of the business strategy. - Pricing shifts from seat-based and subscription models to usage/output-based. - Gross margins become a function of inference efficiency, and will therefore shrink (forget about the 80% margins from SaaS). - And the competitive moat isn't just about the product, but also about how efficiently (and reliably) a business can turn tokens into value. The SaaS era was about selling access, yet the token era is about selling output, and it changes how software companies will be built and valued.

33

3 Comments

AI has hit an infrastructure bottleneck. ICONIQ's latest State of AI report finds that ~23% of revenue now goes to inference, a structural cost that only grows as you scale. At the same time, compute has fragmented across hyperscalers, neocloud providers, and GPU marketplaces, forcing teams to choose between cheaper but unreliable supply or expensive vertically integrated clouds. The result? Months of integration work just to get production workloads running, and even then, performance is unpredictable, and costs spiral. That's the problem Mike Henry and Tim Harris set out to solve with Parasail. I've known Tim since backing him at Qualcomm, and watching him, Mike, and the team lean into that problem early has Touring Capital proud to support their $32 million Series A funding round. Instead of building another cloud provider, they built an orchestration layer that sits on top, aggregates global compute, dynamically optimizes placement, and lets developers deploy production AI endpoints in under five minutes with as little as five lines of code. As with every platform shift, the hardest problems aren’t always the most visible ones. This is one of them, and once you see it, you can’t unsee it: https://lnkd.in/gshgR2tb Samir Kumar Priya Saiprasad Evan Wijaya

55

3 Comments