Steven Xi

Founder, Eastlink Capital. Fmr MD of Riverwood Capital. Focus: AI Infra & CSecurity. Seed to Series A, but flexible. Alum @ChicagoBooth,@USC, Nerd and optimist.

[Yash Patil @ypatil125]

Using open models and inference clouds (which serve open models) is a leading indicator of what is to come.

The advantage of open weights is that you can train, serve, and continually improve your own model. You get control over cost, quality, latency, deployment options, etc.

It's just a matter of time before every company does this for their most important AI workloads.

[Kunal Bhatia @kunalbhatia91]

Superintelligence will be built on Self Improvement.

Today @hexoai, we’re excited to release ‘SIA’ - an open-source Self-Improving AI, to achieve any goal through recursive self improvement.

While trying to solve a problem, SIA doesn't just improve it's abilities by updating it's harness, it updates it's own weights as well.

[Steven Xi @StevenXi07]

@aaref @trajectorylabs @rronak_ @MichaelElabd @QuantumArjun Congrats, Aaref!

[Steven Xi @StevenXi07]

@lqiao @FireworksAI_HQ Amazing! Congrats, Lin!

[Tri Dao @tri_dao]

After some mathematical rewrite, turns out all of transformer is a series of gemm + epilogue. Given a few optimized primitives, LLMs (and novice humans) can write speed-of-light kernels for all transformer ops!

[Steven Xi @StevenXi07]

Congrats and keep it up the good work! @mercor_ai

[Ivan Landabaso @IvanLandabaso]

Mercor went from $1M to $1B+ ARR in 20 months.

The growth playbook behind the 3 youngest self-made billionaires in history 🧵:

[Steven Xi @StevenXi07]

@Sirupsen Congrats, Simon! You are every AI/Data Infra investor’s dream!

[Steven Xi @StevenXi07]

@patrickachase @bernhardsson @akshat_b Congrats, Patrick! Glad to see you are trippling down!

[Steven Xi @StevenXi07]

@ekzhang1 @modal Congrats and thank you for your contribution too, @ekzhang1!

[Steven Xi @StevenXi07]

@itsandrewgao More are coming soon!

[Steven Xi @StevenXi07]

@deedydas Welcome on board, Deedy!

[Steven Xi @StevenXi07]

Congrats to @bernhardsson! In the agentic era, @modal is doing something rare: building a general-purpose compute and AI inference platform, with developer experience as a first principle. Proud and grateful to have backed this team since their Series A. Still early — LFG! 🚀

[Erik Bernhardsson @bernhardsson]

Today we're announcing our Series C funding: $355M at a $4.65B valuation, led by some great investors @generalcatalyst and @Redpoint.

We've had insane growth in the last year, but we're still very early. So proud of the team and what we have built so far!

[Steven Xi @StevenXi07]

@joannezchen @cerebras @andrew Congrats, Joan and friends at Foundation Cap and Cerebras!

[Steven Xi @StevenXi07]

@timshi_ai Amazing! Congrats again!

[Steven Xi @StevenXi07]

I first met Tim when he was still a Ph.D. student at Stanford more than 10 years ago. He made a big impression on me not just because he was a wickedly smart but also forward loooking. Later, he has also proved that he is a great CTO and co-founder at @Cresta. Congrats!

[Nancy Xu @nancycxu]

Congratulations to @Recursive_SI on announcing their $650M raise at a $4.65 billion valuation today, led by @GVteam, @nvidia, @AMD etc -- marking one of the largest formation rounds in history.

@RichardSocher, @_rockt , @josh_tobin_, @jeffclune, @timshi_ai, Alexey Dosovitskiy, @CaimingXiong , @tydsh have build the ultimate team to tackle recursive superintelligence.

Every few decades, a radical new technology accelerates an era of human innovation. I’m increasingly convinced self-improving superintelligence is the fire of the next innovation Renaissance. In a world where the narrative revolves around agents and job automation, what I love about @Recursive_SI’s mission is that they change the narrative -- knowledge discovery and innovation is not a zero-sum game.

Thanks for letting me be a part of the journey. Just a few months ago, the Recursive team was working out of my living room. The sheer momentum and speed they've scaled at to date is a testament to the talent density and conviction of the team.

The team converges the inventors of the biggest breakthroughs in the open-endedness, recursive self-improvement, and frontier models of the last decade: Darwin Godel Machine, Codex, DeepResearch, Genie 1/2/3, ImageNet, Vision Transformers, and more. Alongside the best researchers and leaders from @OpenAI , @GoogleDeepMind , @Meta , and more.

Stacked and just getting started.

[Steven Xi @StevenXi07]

@nancycxu @Recursive_SI @GVteam @nvidia @AMD @RichardSocher @_rockt @josh_tobin_ @jeffclune @timshi_ai Congrats to Tim and the @Recursive_SI team!

[Nathan Lambert @natolambert]

Airbnb CEO Brian Chesky: “We’re relying a lot on Alibaba’s Qwen model. It’s very good. It’s also fast and cheap... We use OpenAI’s latest models, but we typically don’t use them that much in production because there are faster and cheaper models.”

The valley is built on Qwen?

[Steven Xi @StevenXi07]

@BoringBiz_ I enjoyed reading your tweets, but your logic here is totally flawed for obvious reasons (e.g. both can be fairely priced). Excuse me for being blunt.

[Steven Xi @StevenXi07]

Insightful post. Hats off to @chriszeoli

[Chris Zeoli @chriszeoli]

https://t.co/hLt7kX1dhO

[Steven Xi @StevenXi07]

@wolfejosh @LTwolfe Congrats, Josh! You both hit jackpot!