Yi-Jian Ngo

As AI workloads eat up global computing supply, DRAM prices are surging like never before—creating major challenges for teams trying to plan out their data center spend. One of my portfolio companies, Mext, is addressing the largest cost component in the datacenter: server memory (DRAM). From being on the board of Redis for the last twelve years, I’ve learned a lot about the various aspects of this problem and approaches to solving them. Mext found a breakthrough, using new AI techniques, for dramatically reducing the amount of server-DRAM required to run applications, all while maintaining performance. It intelligently manages the server’s memory, keeping hot pages (i.e., those in use by applications) in fast DRAM and offloading cold pages (i.e., those less used) to a much less expensive memory tier (e.g., NVMe Flash). Key to the approach is ensuring cold memory pages that are about to be accessed by an application are back in DRAM before the application needs them or notices that they were gone. This is done without modifications to the application or the OS – so it can run in the cloud or on-premise. Mext thus allows applications to either run using less DRAM or keep their DRAM footprint but do more with it. They're hosting a webinar along with Fred Weber (former CTO of AMD) on Jan 22nd at 10am pacific time. You can register here: https://lnkd.in/gXe9GCeE

153

3 Comments

"Build the event you wish existed" That's what we at Monadical did last week at #NYTW. We wanted an AI Engineers discussion for Engineers. There's always a lot of events out there for VCs to network, or for startups to learn about G2M, but very little on lessons learned from actual engineers in the field. So that's the event we hosted. Big thank yous to our awesome panel, Roy Pereira, Ben Cohen and Corey J. Gallon. Here's the key takeaways and the AI tools we're using. 🚀 All three panelists independently called AI Agents the most transformative LLM application they’ve used. They specifically called out Claude 3.5 Sonnet for its accuracy and reliability. 🪨 We dug into how LLMs are “jagged”, not general. They can be shockingly good at some tasks and completely fail at others. Everyone agreed: good evaluations are critical (and hard.) 🧪 Corey noted how public benchmarks and reality are two different things. Most public evals are saturated or gamed. ♊ Ben emphasized that AI projects are actually two projects: building the tool and building the evaluation process. 🧱 We explored how falling dev costs may impact startup defensibility and labor demand. Roy shared that founders are already shifting strategies in response. ⚒️ In a world of daily AI launches, the panel discussed how they decide what’s worth attention, and what’s just noise. They called out tools like Goose, Aider, Claude Code, and Monadical’s own Cubbi, which helps run agentic workflows safely in dev environments. (links in the comments). CTA: What would you want to hear in an AI Applied Engineering talk you attended?

33

5 Comments

I’m lucky to work alongside colleagues like Richard Fruehauf, who constantly push my thinking. His latest piece unpacks why U.S. robotics and drone manufacturing faces uphill economics, and how public policy can help turn the tide. One key takeaway: 232 tariffs on robotics imports could help reset the economics that are holding back U.S. investment and production. It’s a sharp perspective on how policy can shape the future of advanced tech here at home. Well worth a read: https://lnkd.in/edMPTQPQ Richard Fruehauf, Robert J. Szczerba, Michael Lutzky, Edward Ovando, Karl Sanchack, Herman Herman, Jeff Legault, Phil English, Richard Fruehauf, Ryan Aument, Michael Guido, Sheila Fitzgerald Sterrett, David McCormick

19

2 Comments

Honored and excited to partner with Drew Baglino and the incredible team at Heron Power as they re-imagine the electric grid for the modern world. Power in America is not zero sum. We deserve access to cheap and abundant electricity and we should win the AI race. And we must build the capacity to manufacture critical technology here at home. When faced with existential challenges, we deploy technology against them. And we win.

54

1 Comment

I’m thrilled to share that White Star Capital is backing AMI - Advanced Machine Intelligence in their historic $1.03bn seed round 🚀 - Led by the legendary Yann LeCun and Alex LeBrun. AMI isn't building another LLM, they're building World Models. Why this matters: *Beyond Text: Grounded in JEPA architecture to solve physical-world problems. *Global Powerhouse: A dream team from Meta FAIR, OpenAI, and DeepMind. *Sovereign Tech: A credible European-led alternative to US Big Tech. Excited to be part of this journey along with the rest of the WSC team Eric Martineau-Fortin, Matthieu Lattes, Bérénice Moustial and Victoria Pozzi Rocco Belforti. 🌍✨ https://lnkd.in/eiCKZKSQ #VentureCapital #AI

86

2 Comments

I’ve known David Meister for years and backed him at Sydecar, so I’ve seen his ability to build through complexity. Axiom Trust is taking on one of the most outdated parts of wealth infrastructure, trust administration, by pairing a regulated trust company with AI-native workflows. We're proud to invest again and congrats to the team on the launch!

18

1 Comment

🚀Initiation of Coverage Agility Robotics 🚀 We’ve initiated coverage on Agility Robotics for IPO CLUB’s flagship fund, America2030, within the robotics sector covered by me, with the help of Helen Zhou. Headquartered in Corvallis, Oregon, the company builds bipedal humanoid robots designed to operate in human-centric environments. Its mission is to solve labor shortages and automate dull, dirty, and dangerous tasks across logistics and manufacturing. 🔍Why we like Agility Robotics Commercial first-mover advantage: Achieved the industry’s first revenue-generating humanoid deployment with GXO Logistics, Inc., proving real-world value while competitors remain in pilot mode. Differentiated biomimetic technology: The Digit platform uses dynamic bipedal locomotion, passive dynamics, and a unique backward-leg design for efficient, human-like mobility across stairs, cables, and uneven terrain. Manufacturing scale readiness: The RoboFab facility in Salem— the world’s first dedicated humanoid robot factory—gives Agility a production runway toward 10,000 units/year, a 12–18 month lead over peers. 💡Notable Investors DCVC, Playground Global, WP Global Partners, Amazon, Schaeffler,Sony, SoftBank Investment Advisers, MFV Partners, TDK Ventures ⚔️ Competitive Landscape Primary competitors include Figure, Tesla, Apptronik, Boston Boston Dynamics, and Sanctuary AI. Agility differentiates through logistics-first commercialization, proven deployments, and factory-scale readiness, while others emphasize demos, internal R&D, or general-purpose manipulation capabilities. 🌍 With hundreds of robots planned for production in 2025 and RoboFab scaling toward 10,000 units annually, Agility is emerging as one of the most commercially advanced humanoid companies globally. By focusing on brownfield automation and real-world logistics workflows, it is building the backbone of next-generation warehouse automation and mobile manipulation. #Robotics #Automation #Humanoids #LogisticsTech #DeepTech #VentureCapital #America2030

11

1 Comment

Voice is a super interesting modality right now - maybe the first modality we're seeing move to open source models across a number of scale ups / enterprises. Reliability concerns, high costs, and open source model performance are pushing engineers to do their own fine tuning vs. relying on third-party vendors of proprietary models. Many of these orgs have already been collecting their own first-party data and now with third-party vendors like Extrian, David AI, etc they can train really high quality models. RL has been insanely hyped, but it's been unclear how long it will take scale ups and enterprises to actually lean in. Voice AI might be hitting that inflection point faster than expected.

45

7 Comments

While there has never been a more exciting time to be a founder building dev tooling or next-gen infra, it has also never been less investable at seed/pre-seed. I'm either really missing something or a lot of my peers are lost. As someone who has not just written, but also SHIPPED, about 75k lines of code in the past 6 months I can tell you that the evolution of how to build products has changed as much in the past year as it did in the entirety of 2007-2017. The complete rise and fail of frameworks, platforms, methodologies, etc... paved over and forgotten... that is of course except for the 1 company that gets a 1000x return from a wildly overvalued hyper-scaler or drunken growth stage investor obsessed with compounding at scale. Imagine a world where any seed investor in trends like Openstack, Hadoop, PaaS, etc all took a full loss on their investment. That's what we're looking at right now. I personally know of over a dozen well-funded seed-stage companies building in these spaces, with years of runway, scrambling to get acquired for a return of capital + several million personally while they're still relevant. If you're not seeing this unfold in front of you, you either aren't paying attention or you're satisfied playing the lottery instead of investing.

75

11 Comments

On-Prem is the New Cloud: Why Enterprises are Reversing Stance in the Age of AI Remember when cloud was the battleground for innovation, agility, and scale? Everyone was tracking the % of on-prem compute moving to the cloud and it was growing every year? Well now there’s a reversal. The reality today? On-prem is the new cloud. Here’s why. AI’s Dark Side: Legal Risk and Data Governance Headaches Enterprises are excited about AI, but also uneasy as vendors like OpenAI face intensifying legal pressure. In a high-stakes lawsuit, a judge ordered OpenAI to preserve all ChatGPT chats (even deleted ones), overriding its 30-day deletion policy. Imagine that: you delete sensitive conversations, but they’re still sucked into legal limbo. The precedent is clear: AI prompts and outputs are discoverable records, demanding integration into enterprise ESI (Electronic Stored Information) policies. For large organizations handling sensitive data, the risk is simple: you can’t afford your AI tools to turn into legal liabilities. Especially when global regulations (GDPR, etc.) and internal governance collide with unpredictably broad discovery orders. Sovereignty, Control, Trust Many enterprises still ban ChatGPT or impose strict rules on usage. Even when allowed, token limits are throttled. Enterprises are cautiously letting AI in (as they should). Enterprises need to have control and trust. SAP is leaning into this with its Sovereign Cloud On-Site, deployable on customer premises with SAP-managed infrastructure. It is built to deliver data, operational, technical, and legal sovereignty in one package. Doubt it will be the last. Scaling Pains Talking to founders, three reasons stand out why cloud isn’t scaling with AI: 1/ The rate at which AWS/GCP lets you provision compute and the compute it wants you to provision both falter as AI and AI driven developers deploy a lot more code more frequently than before. So sometimes it’s not about data governance, but actually about performance and functionality. 2/ Shadow AI is rampant—employees paste sensitive info into models despite restrictions—creating demand for better permissioning tools. 3/ Data leakage remains a live risk, like the August 2025 Grok incident where hundreds of thousands of private chats became publicly accessible and indexed. One thing is for sure - founders selling into the enterprise are scaling faster if they offer the ability to run on-prem. The Future is Hybrid Don’t get me wrong - cloud is not going anywhere. But neither is on prem. A few years agoI thought we were headed for a full cloud shift, but now I think the future is truly hybrid.

34

4 Comments

Don't Build a Better Wheat Farm" - Why Defensibility Stakes Are Higher in Deep Tech Just published a new piece on my "Ignore the Confusion" blog, building on thoughtful insights from Eric Ver Ploeg at Tunitas Ventures about startup defensibility. Eric's core thesis: Too many startups pitch like wheat farmers - "huge TAM, slow incumbents, growing market, domain expertise" - but fail to think through long-term defensibility until it's too late. From a deep tech perspective, the stakes are even higher: ** Unlike software, deep tech founders must commit to defensibility strategies from day one - their funding depends on it ** Patent vs. trade secret decisions are often difficult to reverse and shape your entire competitive strategy ** Even "picks and shovels" providers (the tools that make industries more efficient) become commodities without proper moats The key insight that resonates: Defensibility can't be retrofitted. Whether you're building software or deep tech, your moat must be architected into the business model from the start. Thanks to Eric Ver Ploeg for sharing these insights on startup strategy and letting me build on his framework from a deep tech lens. Read the full post: https://lnkd.in/dEj_iF-Q #DeepTech #StartupStrategy #Defensibility #VentureCapital #Innovation

21

2 Comments

With RL techniques like GRPO, it’s become easy to fine-tune small models to outperform foundation models on vertical tasks. We’re open sourcing Osmosis-Apply-1.7B: a small model that merges code (similar to Cursor’s instant apply) better than foundation models. Links in comments! Using foundation models for high specificity, low complexity tasks like code merging can be overkill - instead, it’s better to use a specialized model instead for higher performance and latency. We fine-tuned the model on 100K real-world code commits where the model needs to merge a code edit snippet into an original code snippet. On the test dataset (subset of CommitPackFT), foundation model performance ranged between 0.77-0.93 reward score. In comparison, Osmosis-Apply-1.7B achieved a 0.98 reward score - while also being 3X-5X more affordable than the next cheaper model AND ~10X faster! Why this matters: the best AI agents use multiple models to ensure task specialization and reliability at scale - on top of cost and latency benefits as well. Links to the model and full write-up (with reward function, hyperparameters, etc.) are in the comments - reach out if you’re interested in using reinforcement learning!

240

18 Comments