After 10 years supporting the investment team and portfolio companies as an executive-in-residence and operating partner at Battery Ventures, I am excited about my new role as a partner focused on frontier tech investing in Battery Ventures XV, our firm’s new, $3.25 billion fund.

My main focus will be on the AI models, but I will also be spending time on quantum computing and robotics/physical intelligence.

The industry has made great progress on AI models, but in my view we are still very early in the journey. One example: How many billions of dollars did it take OpenAI to train a model that tells me to walk to the carwash if it’s nearby rather than driving. Maybe when they get to $1 trillion of training resources the technology will anticipate my follow-up question—how, if I walk, will I actually be able to wash my car? Or maybe it won’t. While this level of contextual understanding may be harder than it superficially appears it doesn’t seem like it should be even close to a trillion-dollar problem.

I believe there must be a better way. New structures and algorithms will be needed to continue advancing the state of the art with reasonable efficiency.

As a formerly decent tournament Go player (1-2 dan), I was stunned roughly a decade ago to see computer AlphaGo beat Lee Sedol, a top professional Go player, and even more impressed to see AlphaZero teach itself Go with no human guidance. The fact that a moderate resource investment could produce superhuman reasoning in a deeply complex game that had proven impervious to traditional methods showed me the promise of AI. The ability of current models in coding will be a revolution in software development, but much work remains.

How will we get from where we are today to superintelligence across a broad range of fields? My bet is that most of that progress will come from algorithmic breakthroughs, with only incremental gains coming from further scaling of training investment. I am particularly excited about opportunities in three areas:

– New LLM architectures that scale more efficiently
– New approaches to memory for AI systems
– New approaches to the software development lifecycle that build on the progress that has been made in AI driven coding.

Stay tuned for a companion post that dives into why I am excited about these themes.

So, why the change in my role at Battery? Why not just continue as an operating partner, helping the investment team find the next generation of exciting companies and helping founders grow them? I would summarize my decision with one word: accountability.

As CEO at MongoDB, and before that in a variety of executive roles, I was accountable for a company’s or a business unit’s results. For the last decade working at Battery, I was in a support role, accountable for the quality of my work but not the outcome.

I believe accountability sharpens focus. Meeting entrepreneurs with an objective of delivering an assessment to my partners is different from meeting entrepreneurs to decide if this will be one of the handful of investments I will make in any given year.

During the course of 2025, I became more active around new investments. I was part of the Battery team that invested in the Databricks* Series J round (having first invested in the series D); ClickHouse* Series C; Reflection* Series B; Fundamental* seed and Series A; and Quantum Art* Series A. These investments gave me confidence that I would be able to find investments in Battery XV that I am passionate about.

Nothing could excite me more than waking up each morning and feeling both the opportunity and responsibility to find and partner with the founders that are building what I believe will be the most important technologies of my lifetime. I am honored that my partners have trusted me with this responsibility and am determined to live up to their trust, and to the trust of the founders with whom I partner.

I look forward to sharing more about what I am seeing in the market and the founders I am partnering with in Battery XV.