This is the biggest single investment we’ve made as a fund to date.

$300,000 split across three different Bittensor subnets, all operated by the same team – Macrocosmos.

Why?

Because in an ecosystem filled with brilliant ideas and half-baked follow-through, Macrocosmos consistently ships. They don’t just publish diagrams. They launch subnets, onboard users, write docs, push code, and build actual products that solve real-world problems.

We’ve been watching Will Squires and Steffen Cruz for a while. The way they lead, the people they attract, the clarity of thought behind each subnet – it’s rare.

We’re proud to back them, and we want to tell you why.

Subnet 1 – Apex
Apex is the proving ground for agentic AI – where models compete in live, adversarial tests to see which ones can actually reason under pressure. Think of it as a competitive arena where LLMs are judged not by benchmarks, but by how useful their answers are in the moment.

This subnet powers high-quality inference and rewards the best outputs – helping weed out hallucinations and strengthen decentralized AI models across the protocol. Over time, we believe Apex will power enterprise interfaces and paid APIs, serving up intelligent agents for coding, web search, and customer support.

Subnet 13 – Data Universe
Data Universe is the real-time data layer for Bittensor. It scrapes, stores, and organizes massive volumes of content from the open web – like social sentiment, financial data, and live news feeds – and makes that available to other subnets and users.

What makes it special is the structured query system and how it incentivizes freshness, uniqueness, and high-value sources. Early users include funds and traders tracking Reddit and Twitter buzz, fintech teams training fraud models, and sports prediction tools calibrating odds with live sentiment.

It’s also one of the most commercially advanced subnets in the ecosystem. Through streaming APIs, historical datasets, and structured dashboards, it’s already onboarding early users who want real-time social signal at scale.

Subnet 25 – Mainframe
Mainframe is Bittensor’s science stack – a decentralized supercomputer built for molecular simulations.

Its flagship use case is protein folding – a notoriously complex, expensive, and time-consuming task in pharmaceutical R&D. Mainframe lets researchers simulate those experiments in hours instead of weeks, using decentralized GPU miners around the world.

That means drug discovery can move faster, cheaper, and with less risk. Billion-dollar pharma companies spend over $100B per year on R&D. Mainframe offers a shortcut – a way to triage ideas, reduce wet lab failures, and accelerate time-to-market.

The road to revenue is clear: folding-as-a-service via APIs, licensing simulation data, and deeper integrations with biotech players like Rowan Scientific.

Together, these three subnets – Apex, Data Universe, and Mainframe – represent three distinct layers of AI infrastructure:

• Apex strengthens intelligence through competition
• Data Universe structures and supplies real-time signal
• Mainframe turns decentralized compute into tangible scientific value

And Macrocosmos is at the centre of all of them.

Siam Kidd, DSV CIO: “Macrocosmos is what conviction looks like. They’ve taken the hard route – open-source, distributed, deeply integrated – and are producing outputs the entire protocol benefits from. This isn’t an investment. It’s a partnership.”

About DSV

DSV is the only hedge fund in the world fully focused on Bittensor. Our edge isn’t traditional diversification – it’s subnet intelligence.

With over 100 subnets in the ecosystem, all the variety we need is already baked into the protocol layer. We hold the base currency (TAO), we trade subnets, and we invest strategically in high-calibre teams building real products – typically through OTC deals like this one.

More importantly, we’re not watching from the sidelines. We’re inside the ecosystem – talking to founders, attending the events, joining private calls. We see the best deals before they hit the public radar, and that gives us an unfair advantage.

We’re looking for investors who believe the future of AI will be distributed, not centralised – and who want early exposure to the asymmetric upside Bittensor unlocks.

We hold the protocol layer (TAO), we trade subnets, and we do strategic investments in high-calibre teams who are building useful, real-world products on top of the protocol – typically through OTC deals like this one.

We’re not tourists. We’re embedded in the ecosystem, in constant dialogue with builders, attending the events, joining the private calls. We see the best deals before they hit the public radar – and that gives us a true asymmetric advantage.

We’re looking for investors who believe that the future of AI will be distributed, not centralised – and who want exposure to the exponential upside that Bittensor unlocks.

If that’s you – we should talk.