The New York Times recently highlighted the struggles of startup founders attempting to access computing power for their artificial intelligence projects. Due to the competition from big companies and even nations, these founders are having difficulty obtaining the needed graphics processing units (GPUs). One founder of Index Ventures commented on the situation, describing the GPUs as “rare earth metal.”
It contacted data center operators around the world, offering to pay them to “mine” GPUs for its portfolio companies. Founders are attempting various strategies to acquire chips, such as asking favors from friends at large equipment vendors that may have GPUs to spare, and exploring the U.S. government’s Access program. Index Ventures, a global investor, proposed another idea to the outlet. They reached out to data center operators worldwide, offering to pay them to mine GPUs for their portfolio companies.
Oracle has struck a deal with a company to provide its founders with Nvidia’s H100 chips and Nvidia’s A100 chips to prevent its portfolio companies from facing a shortage.
We spoke with Erin Price-Wright, a Bay Area-based partner with Index ventures who focuses on enterprise software and AI, earlier today. Price-Wright, who was the head of product for Palantir’s data analytics and machine learning platform before joining the venture firm in 2019, shared insights on the arrangement that other venture firms are trying to replicate. Here are lightly edited excerpts from our chat; you can listen to the full conversation here.
Discussing partnership with Oracle
We’re partnering with Oracle to make GPUs more accessible to early-stage AI companies. We’re the only company that has the ability to access Oracle’s GPU capacity and make it available on the spot.
Trying to get GPUs can take three months to a year, making it hard to get started.
Early-stage companies often find it difficult to understand what their workloads will look like. To help remove this barrier of access, we are partnering with Oracle to provide GPUs to our earliest-stage portfolio companies. Our ultimate goal is to help these companies transition to their own cluster and focus on what matters from day one.
We don’t offer massive GPU clusters to our companies, but we want to give them a head start so they can begin building quickly and level the playing field.
We can actually help these companies come together with the right people to make sure that they can get access to the right technology. We wanted to ensure that entrepreneurs tackling concrete business issues did not have to alter their business model, presentation, or fundraising approach to gain access to GPUs. We observed this pattern repeatedly with early-stage companies, and realized Index ventures could provide leverage by connecting them with the right people to obtain the right technology. Thus, the deal was born.
We leverage our market position, relationships, and ability to aggregate demand from multiple companies to offer value-added services to our founders.
Do we give Oracle a stake in the startups?
Index makes a pre-commitment to pay the cloud bill on behalf of our startups and Oracle manages the cluster. Oracle has been an excellent partner, and our companies can access the GPU cluster free of charge.
Did you have to discuss the payment of the cloud bill with your investors since it is not a typical venture firm move?
I will not share many details about how the agreement works yet. Is this an exclusive relationship that prevents other venture firms from doing the same thing? No, there is no exclusive relationship with Index. One benefit Oracle gets from this is the opportunity to meet the next generation of outstanding companies as soon as possible.
We actively help our companies navigate the process of signing their own dedicated cloud deal using our GPU cluster. The goal is for them to develop relationships with Oracle, AWS, and other large cloud providers and sign their own dedicated contract, rather than relying on us in perpetuity.
We’ve achieved this with Cohere, as Oracle and Nvidia have both backed them. These two companies are highly sought after right now and it’s great that we’ve managed to get them involved in our portfolio.
Index Ventures Investments
VCs are investing a lot of money in generative AI, and Index ventures has at least 20 portfolio companies in the AI/ML sector. These include Cohere, which has raised $445 million, and Mistral AI in France, which recently raised a substantial seed round.
I believe we are quickly entering a period of decreased enthusiasm, particularly regarding large investments from traditional venture capitalists. There is still a wide gap between the potential of technology and what is necessary for it to be applied and beneficial in the business world.
We need to fill a large infrastructure gap, but it won’t happen quickly. It will take some time.
I remain highly enthusiastic about the potential of the core technology and its ability to revolutionize the world. Over the next year, however, businesses will have to work hard to identify the return on investment, prioritize the use cases and develop more than just a few prototype demonstration applications.
We will begin to witness the infrastructure that will enable these use cases to be supported at scale. As an investor, how do you ensure your artificial intelligence companies do not overlap? Is it any more difficult than when dealing with traditional startups? I do not believe it is significantly different from how we consider competition in other areas. Everyone portrays AI as a unique category.
But they all use SaaS. In two years, artificial intelligence will be the beating heart of every piece of software. No code, software, application, or website will be exempt from AI as a core component. It’s almost like how every SaaS company uses SaaS, but they each have their own unique characteristics.
Every SaaS company has a database, a front end, and an interaction between the two. AI is comparable to a database in terms of its core function in developing software. As we are still in the early stages of the market, companies are still exploring the possibilities of these tools and which specific problems they should address.
I view the concept of competition in the artificial intelligence landscape much like any other sector. AI is often portrayed as a distinct category, yet when I project just a couple of years ahead, let alone five or a decade, every software we engage with will be underpinned by artificial intelligence. No single piece of code, software, application, or visited website will exclude artificial intelligence as an integral element. It’s comparable to the evolution of SaaS: each SaaS company varies, yet they all incorporate databases and front ends, interacting cohesively. AI parallels a database in its fundamental role as a cornerstone of software development.
Read More: About AI News
While the market is in its nascent stages, fluctuations and adaptations will occur as companies determine how to leverage these tools for specific challenges. This dynamic is akin to our approach in traditional SaaS investment, from my perspective.
