Joined July 2009
1,354 Photos and videos
Pinned Post
Today we're announcing over $9B in capital for a new family of funds — our largest raise in 25 years. AI is rewriting the rules. We're backing the founders building what comes next.
54
40
561
121,448
Can space data centers actually work? @guruchahal says the real bottleneck is not radiation or cooling, it’s launch cost. "We've been sending compute into space for 17 years. Every satellite that is up in space is running a CPU, tapping into the sun's power, removing heat, and handling radiation." "The question of can you run a GPU or a CPU in a satellite up in space is a foregone conclusion. You obviously can." "The real question is: can you do it in an economically feasible way?" "If it costs zero to put compute up in space, space would be significantly cheaper because power is free and cooling is free." "The big part is what SpaceX and other launch companies are able to do in terms of bringing the launch cost down." From Lightwork.
NEW EPISODE: Midjourney’s body-scanning spa, ChatGPT slips below 50% market share, two big DeepMind departures @Lightspeedvp Partner @guruchahal breaks down AI data centers, layer by layer, and where we see innovation happening 3:04 Midjourney Medical: A Spa That Scans Your Body 11:26 The Assistant Race: ChatGPT Slips Below 50% 15:59 DeepMind Departures And The AI Talent War 19:57 Google x A24: AI Goes To Hollywood 28:01 Debunking AI Data Center Myths 33:20 How AI Is Making Data Centers More Efficient 37:26 The AI Data Center Stack 49:36 Can We Solve the Biggest Challenges of Space Data Centers? @Machiz
2
5
38
7,916
Everyone is talking about data centers. But what actually are they? @guruchahal gives the simplest explanation: "Data centers are nothing new. They've been the anchor and a critical part of our economy for over 30 years." "Everything we do in technology comes from a data center. You watch Netflix, that's a data center. You do a bank transaction, that's a data center. You buy something from Amazon, that's a data center." "They're just a collection of computers sitting in a large building, serving up our websites and technology and applications." From Lightwork.
NEW EPISODE: Midjourney’s body-scanning spa, ChatGPT slips below 50% market share, two big DeepMind departures @Lightspeedvp Partner @guruchahal breaks down AI data centers, layer by layer, and where we see innovation happening 3:04 Midjourney Medical: A Spa That Scans Your Body 11:26 The Assistant Race: ChatGPT Slips Below 50% 15:59 DeepMind Departures And The AI Talent War 19:57 Google x A24: AI Goes To Hollywood 28:01 Debunking AI Data Center Myths 33:20 How AI Is Making Data Centers More Efficient 37:26 The AI Data Center Stack 49:36 Can We Solve the Biggest Challenges of Space Data Centers? @Machiz
5
5
16
4,847
.@bsomaia on AI valuations: “Are we in a super cycle? Yes.” “Are we in a very rich valuation environment? Yes.” “Will there be investments that despite this valuation environment will be extremely rewarding? Yes.” "Will there be investments that turn out to lose money? Yes." "All of those things can be true." From Startup Policy Forum with @ShwetaRKohli.
3
4
27
2,903
The team at @unconvAI has released their first model, Un-0! Un-0 is an image-generation model powered by a simulated system of coupled oscillators. Instead of the massive arithmetic operations that power most AI, the model uses simulated physics. Thousands of virtual oscillators influence each other, sync up, and settle into patterns. Feed it an input, let the oscillators interact, and the resulting pattern is the output. On ImageNet 64×64, Un-0 achieves an FID score of 6.74, matching the quality of leading conventional image generation methods when they were first published. The company has stated that, to their knowledge, this is the most capable image generator to date that uses a simulation of a physical dynamical system. Unconventional AI believes energy efficiency demands a fundamentally different computer, one where physics does the computing. We believe this is a meaningful milestone in their goals of achieving ~1,000x energy-efficiency gains in modern AI. Congratulations to @NaveengRao, @MeeLan_Lee, @sachour__, @MCarbin, and the entire Unconventional AI team. We are incredibly excited for what’s ahead. CC: @guruchahal, @ravi_lsvp, and @buckymoore
Introducing our first model, Un-0! We trained an image generator powered by a backbone of coupled oscillators in place of a more traditional conventional neural network.
1
3
18
3,170
A big congratulations to the @Arena team on hitting $100M ARR! This milestone announcement comes just eight months after the launch of their evaluation product. Arena hosts millions of users who can work in real time with AI agents, testing and comparing different AI models. One primary feature is their AI model performance leaderboard, which has been generated from over 10 million user evaluations. Incredible milestone @ml_angelopoulos, @infwinston, @istoica05, and team. We’re proud to see Arena continuing to develop and advance the frontier of AI through community-driven evaluation. Learn more in the post below.
Arena has crossed $100M in annualized revenue run rate, eight months after launching our evaluation product. With our recent release of Agent Mode, millions of users on Arena are doing real work with agents, from coding to document analysis, in long-running, multi-turn sessions with hundreds of tool calls. Arena now evaluates objective criteria like task completion rates, hallucination rates, and more, far beyond our original human preference voting model. This expansion has taken us from a student project at Berkeley to one of the fastest growing companies in history. Go Bears! 🐻 Our core thesis is simple: to align AI with human values, we must directly measure its impact on people in the real world. Today's milestone is proof that Arena’s platform is the de-facto standard for post-deployment evaluation of AI.
1
15
2,669
When intelligence is abundant, opportunities are omnipresent. Lightspeed’s @bsomaia spoke with @ShwetaRKohli, President & CEO of Startup Policy Forum India, to cover what he believes the AI super cycle could mean for India’s digital economy. India already has the entrepreneurial spirit. Pair that with AI, and India could be positioning itself on the cusp of a large-scale transformation. This conversation goes beyond the headlines, unpacking the questions defining the AI era, how venture has changed, and why we believe the next generation of Indian consumer leaders could be AI-native. Check out the conversation below.
‼️🎬 Bejul Somaia joined Lightspeed in 2008 to build the firm's India practice from the ground up. Today, he sits at the centre of the firm’s global portfolio with bets spanning @AnthropicAI, @MistralAI, @emergentlabs, @SarvamAI and many more. We visited him at Lightspeed's Sand Hill Road headquarters for a wide-ranging conversation on the AI super cycle, the India opportunity, and why venture capital now feels “wildly different” making investing “very hard” in an AI era. Catch the latest episode in Startup Policy Forum’s AI in Action (Global Edition) series. Excerpts below and link to full conversation here 👇 youtu.be/VT97Aqqyk1w?is=u96E… @bsomaia @lightspeedvp @LightspeedIndia @SeekingN0rth @dkhare @MohapatraHemant @rahultaneja
1
1
9
2,662
We’re excited to continue backing @straikerai for their $64M Series A. As enterprises apply agents to their workforces, Straiker helps ensure secure deployment with comprehensive discovery, adversarial testing, and runtime protection. When we first partnered with founders @ankurdotshah and Sreenath Kurupati, security agents were still early in enterprise adoption. Now that they’ve become a critical element, the two have grown Straiker to account for that shift. Congratulations to the Straiker team on the raise. It has been exciting to watch the company grow since we first invested in their Seed in 2024. CC: @guruchahal, @buckymoore, @arifj
Straiker has raised a $64M Series A, bringing total funding to $85M, to secure the agentic workforce. AI agents are becoming the fastest-growing workforce in the enterprise. They are writing code, connecting to tools, accessing data, and taking action across real systems. Some will be useful. Some will be weaponized. 𝑻𝑯𝑨𝑻 𝑰𝑺 𝑾𝑯𝒀 𝑺𝑻𝑹𝑨𝑰𝑲𝑬𝑹 𝑬𝑿𝑰𝑺𝑻𝑺. Thank you to our customers, partners, team, and investors for helping us reach this milestone. We’re just getting started.
2
1
12
3,954
Is ChatGPT really drinking a bottle of water every time you prompt it? Lightspeed partner @guruchahal says "the only thing drinking a bottle of water around here is me, not ChatGPT." He explains: "The next generation of AI data centers are closed-loop systems. It's not like there's a river flowing next to the data center and hot water comes out." "Once the data center is up and running, the water just gets recirculated within the system." "We are rapidly approaching a world where the water impact, which is already negligible, is approaching zero." From Lightwork
AI uses so little water it’s almost comical that it even became a topic
4
7
29
7,103
Enterprise tools don’t always translate to small businesses. @getpiehq is coming out of stealth with a growth platform seeking to help SMB owners. We’re proud to have led Pie’s $19.5M Series A as they come out of stealth, bringing their total funding to $23.7M The co-founders of Pie, @akhilmantrip and Syed Ali, have been working with small businesses for a while. They noticed that many had huge growth potential but lacked the tools to unlock it. Enter Pie. Pie is an agentic growth platform that gets the word out on websites people actually use. It runs the business’s ads across all the places that potential customers might look, and it aims them at the customers ready to buy, not just whoever is cheapest to reach. This round of funding coincides with scaling of Pie’s vertical SaaS integrations, expand its product suite across growth, AI search, and AI front desk Congratulations to the team at Pie on the raise.
2
3
13
3,294
"[India's] government has generally been so embracing of entrepreneurship and of technology and what technology can do. That mindset is more important than anything else." @bsomaia on what India's government can do to help accelerate transformative technology: "Ultimately, all of this is about animal spirits." "We don't need to figure out the how. Let's lay out a vision for what this could mean for [India] and for different parts of the economy." "Entrepreneurs will go and find the opportunity. We've seen that happen. We'll see it happen again." From Bejul's interview with Startup Policy Forum India
1
2
17
2,781
Lightspeed reposted
it feels like ai researchers are getting traded like athletes. this week on Lightwork, @clairejyz and i unpack Meta’s AI reorg and morale issues, ChatGPT slipping below 50% assistant market share for the first time, two major DeepMind researchers heading to Anthropic and OpenAI, and why Midjourney’s move into healthcare is more significant than it looks. we also get into Google DeepMind’s partnership with A24, where i seem to be in the unpopular camp of people who are actually thrilled. thank you to @Lightspeedvp partner @guruchahal for joining us. we pulled back the curtain on AI data centers and why the energy conversation is more nuanced than the headlines suggest. the stat that stayed with me: Guru framed California almond farming as consuming in roughly 4 days what U.S. data centers consume in water in an entire year. tune in below for our first shoot on the road. and we’ll see you again next week.
NEW EPISODE: Midjourney’s body-scanning spa, ChatGPT slips below 50% market share, two big DeepMind departures @Lightspeedvp Partner @guruchahal breaks down AI data centers, layer by layer, and where we see innovation happening 3:04 Midjourney Medical: A Spa That Scans Your Body 11:26 The Assistant Race: ChatGPT Slips Below 50% 15:59 DeepMind Departures And The AI Talent War 19:57 Google x A24: AI Goes To Hollywood 28:01 Debunking AI Data Center Myths 33:20 How AI Is Making Data Centers More Efficient 37:26 The AI Data Center Stack 49:36 Can We Solve the Biggest Challenges of Space Data Centers? @Machiz
1
1
12
2,121
Lightspeed reposted
A fact that some people are forgetting: Data centers aren't a new or novel concept. They’ve been used to power the internet for at least 30 years now. This week, I joined @Lightspeedvp’s new show Lightwork, hosted by @clairejyz and @Machiz, to talk about the realities of the data center - debunking some common environmental fears - and how tech companies are responding to climate pushback. About 10 years ago Google pointed its own AI at its own data centers and made them 40% more efficient. As AI evolves, I predict that the self-fulfilling cycle of innovation in AI data centers will compound to 100-1,000x higher efficiency. Thank you to Claire and Josh for having me on the podcast this week! You can watch the full episode below. Skip to 23:04 to hear our discussion on AI data centers.
NEW EPISODE: Midjourney’s body-scanning spa, ChatGPT slips below 50% market share, two big DeepMind departures @Lightspeedvp Partner @guruchahal breaks down AI data centers, layer by layer, and where we see innovation happening 3:04 Midjourney Medical: A Spa That Scans Your Body 11:26 The Assistant Race: ChatGPT Slips Below 50% 15:59 DeepMind Departures And The AI Talent War 19:57 Google x A24: AI Goes To Hollywood 28:01 Debunking AI Data Center Myths 33:20 How AI Is Making Data Centers More Efficient 37:26 The AI Data Center Stack 49:36 Can We Solve the Biggest Challenges of Space Data Centers? @Machiz
2
18
3,928
NEW EPISODE: AI engineering talent is fractured, and companies are making shocking bets.
NEW EPISODE: Midjourney’s body-scanning spa, ChatGPT slips below 50% market share, two big DeepMind departures @Lightspeedvp Partner @guruchahal breaks down AI data centers, layer by layer, and where we see innovation happening 3:04 Midjourney Medical: A Spa That Scans Your Body 11:26 The Assistant Race: ChatGPT Slips Below 50% 15:59 DeepMind Departures And The AI Talent War 19:57 Google x A24: AI Goes To Hollywood 28:01 Debunking AI Data Center Myths 33:20 How AI Is Making Data Centers More Efficient 37:26 The AI Data Center Stack 49:36 Can We Solve the Biggest Challenges of Space Data Centers? @Machiz
1
11
3,306