Venture Fund Offering Compute for Equity in AI Startups

The dedicated fund uniquely offers compute in exchange for equity, solving critical capital access issues for AI startups.

### Innovative Venture Fund Structure: Offering Compute for Equity in AI Startups
* 💼  The venture fund aims to leverage **extensive experience in AML investing**, having previously backed companies like **CORE and Cohere**, to create a unique opportunity for **capital investment in the AI space**.
* 🔗  This fund provides portfolio companies with **dedicated GPU compute resources** in exchange for equity, addressing a common challenge where **access to compute is often a barrier to raising capital**.
* 💰  The valuation process for equity in exchange for compute is negotiated, ensuring that the arrangement **solves significant operational issues** for companies needing computational resources.
* 📊  Concerns about overvaluation in the Gen AI sector are acknowledged, emphasizing the importance of investing in companies with **strong management, technology, and market position**.
* 🏦  The dedicated fund structure is beneficial as it specifically offers **compute resources**, distinguishing it from other investment vehicles that typically span **public and private equities**.
* 🤝  The partnership with CORE has evolved over the years, allowing the fund to **secure necessary compute resources** to meet **growing demand from portfolio companies**.
* 💻  Depending on the nature of the company, compute expenses can range significantly, with application developers potentially allocating around **25% of their capital raise** to compute, while large model developers might spend **75% or more**.
* 🔍  This innovative approach not only provides **essential resources to startups** but also positions the fund strategically within the **competitive AI investment landscape**.

You've already backed some well-known names in the VC space. I'm thinking cohere. I'm thinking of core weave. Why go more focused and have a fund? More than $200 million? One of that? Yes. Well, thank you for having me on. I think because of that, because of our experience in AML investing, going back to 2021 with CORE, we've originally and now also cohere, we feel like we have the relationships and the expertise in the space as well as just the time and unique opportunity it is with all the capital going in. And finally, it's because we have a really unique product offering both the support of Magnetar and also the dedicated compute on the core. We've cluster that we think just provides unique value to potential portfolio companies. And so your fund has a really interesting structure where it's providing your portfolio companies with compute in exchange for equity. Tell us how that works and how you determine how much computers worth, how

much equity. Yes. So we've contracted compute that's dedicated GPUs for this purpose with CORE. We've and so potential portfolio companies will get access to the compute in exchange for equity. So like any other investment, it'd be a negotiated process on the value. But the important thing is that it solves a lot of problems for the company. Like oftentimes there's this chicken and egg issue where it's hard to contract compute if you don't have the capital, it's hard to raise capital if you don't have access to the compute. So potentially we're solving that problem by providing a direct exchange of the equity for the compute on core weaves. What we believe is, you know, the top performing high-performance compute cluster. And so how are valuations looking when it comes to Gen AI companies? Is there concern that with the hype in the space that companies might be either overvalued or equity might be expensive? Yes.

Well, I mean, at least we've had a little bit of pause on that, but I think it's hard to generalize. You're part of our investment process is to focus on companies that have both internally, the right management team, the right technology, the right moat, but are also in the right parts of the AI stack, whether that's applications or verticals where it's hard to separate the pieces so that we can be confident they're creating value. Jim I know a lot of firms that want to hold both private and public equities, right? And they try and work out the best mechanism or vehicle to do that. Then you have like the mutual funds who are increasingly high up the cap table for growth stage names. With that in mind, why the venture vehicle? Specifically, what's the benefit of launching a dedicated fund like that? Well, I think it's the structure and the unique value proposition.

So the reason it's a dedicated fund is because the fund is specifically offering compute in exchange for equity. In our other funds, we've historically invested across the cap structure and public equities, private equities, structured debt. But here, like because of the unique value proposition in the way we've contracted the GPIO cluster, it makes sense to offer it as a dedicated fund. It's so fascinating that almost you're going up against, well, this is how us Amazon has got in with a lot of these strategic investments. I'm sure they're offering the compute the Microsoft. What is it? What's the relationship you now have with CORE? We've to be able to ascertain and ensure that you get that compute. Is that kind of a deal that you struck with them And and they're happy, of course, to have yet more and more demand coming for you, I'm sure. So we've been close partners of core we for a number of years now, both in

providing them capital and just generally providing Magnetar, Magnetar resources to help them grow their business. And so this just was sort of the next step in the evolution of our relationship where we felt like between us, we're in a position to offer this product and we were able to work with them on contracting the compute and arrange the fund such that we can deliver that compute in exchange for equity. And, you know, compete in hot demand. But give us some context. If a fund if a company is doing a funding round, roughly how much of that money raised could be spent on compute? So it really depends on the company and how intensive their compute needs are. So if it's an application developing firm, it may be something on the order of 25% of a cap raise might go to compute if they're developing some particular application, if it's, for example, an L am company that's really

developing large models, then you know, as much as 75% or more of a particular cap raise might be allocated to high-performance compute.

Hedge Fund Magnetar Capital Starts AI Venture Fund

Welcome to the Bloomberg Technology YouTube channel. Subscribe for original series and mini-docs on game-changing technologies, profiles on innovators, live conversations with experts, and breaking technology news & analysis from Bloomberg TV.

Bloomberg Technology

Unique Fund Structure for Gen AI Companies

The fund, with over $200 million, leverages experience in AML investing and has a unique product offering of dedicated compute in exchange for equity.

This structure aims to solve the common chicken and egg problem of accessing capital and compute.

The fund's approach addresses concerns of potential overvaluation in the Gen AI space by providing a direct exchange of equity for compute resources.

The investment process focuses on companies with the right management, technology, and positioning in the AI stack.

The dedicated fund offers compute in exchange for equity, providing a unique value proposition.

The fund's structure is designed to meet the specific demand for compute, with a close partnership established with CORE for mutual benefit.

Supporting Business Growth through Capital and Resources

Magnetar is providing capital and resources to help businesses grow, evolving their relationship to offer compute resources in exchange for equity.

The amount of funding raised by a company that goes towards compute varies, with application development firms potentially spending around 25% of their capital raise on compute needs.

The intensity of a company's compute requirements plays a significant role in determining the allocation of funds.

Allocating Capital for High-Performance Computing

A significant portion, up to 75% or more, of capital raises for developing large models is allocated to high-performance computing.

This emphasis on computing power is crucial for the success of these models.

The trend highlights the growing importance of advanced technology in model development.