Beyond the Hype Understanding the AI Startup Ownership Structure and How Founders Balance Equity with Massive Compute Costs

You may have seen some staggering value. New start-up companies, recently formed just 2 years ago, are receiving valuations in excess of $5 Billion, and new seed rounds of funding are becoming larger than the Series C round of most companies. While we in the technology sector are used to substantial figures in the tech marketplace, the technology now being developed using Artificial Intelligence (AI) has driven valuations to an entirely different level. In the past, if you were creating a software company, your largest expense would be your laptop and some coffee if that was true. Today, if you are building a Foundation Model, you are burning millions of dollars each month just to keep the servers operating. This huge demand for capital has changed the entire model for how AI startup ownership structure as well as how these entities are governed and structured.

Why the AI startup ownership structure is built on “compute”

Historically, Silicon Valley entrepreneurs were advised to remain lean. They hired a few engineers, developed a Minimal Viable Product (MVP), and attempted to gain traction before raising the next large round. However, there is no more ‘lean’ definition when developing artificial intelligence systems. Once development begins, there will likely be an enormous ongoing expense associated with paying either Nvidia or the cloud service business that provides your computing resources. This economic reality subsequently drives the “AI startup ownership structure.” As AI start-up companies require a great deal of cash prior to developing an MVP or launching product-market strategies. They typically prefer to surrender more equity at an earlier stage than most traditional venture-backed companies.

By the time founders reach Series A funding, they may only own between 40-50% of their company. This would appear to be a warning sign in traditional SaaS businesses, but in an AI company it represents just the cost of doing business. Starting a restaurant is similar to beginning a new artificial intelligence enterprise. Or opening a restaurant cannot be accomplished with just a toaster. You must have an entire commercial kitchen with the finest location and the very best chefs when the restaurant opens. Therefore, in order to support so many upfront costs, more investors must be involved as early as possible.

The balance of power in the AI startup governance model

When companies with very deep pockets like Google and Microsoft are making billion-dollar investments into AI start-ups, one would assume that they would want 100% ownership of the company. However, the founders of AI start-ups are very protective of their “vision” and do not want any big corporations to potentially lead AI into generating profits at the cost of being safe and ethical. This has led to the emergence of a distinctive governance model for AI start-ups, which is characterized by unusual structures like “dual-class shares”. And which allow founders to retain roughly 60% of the voting power of the start-up even if they only own 10% of the actual equity of their company. The AI start-up board structure also tends to be a unique mix of academics, safety experts, and large-scale venture capitalists (“VC” s ) .

OpenAI can be viewed as a good example of this type of unusual structure – they originally started as a non-profit and have since formed a “capped-profit” subsidiary. Although many AI start-ups will not pursue set up this way, the current structure of AI companies does not consistently adhere to a simple distribution of 80% to founders and 20% to investors. The trend continues to be a situation where founders maintain control while investors continue to support start-ups financially.

AI Startup Ownership Structure: How Investor Relations Deal with “Strategic Money”

When one examines the AI startup ownership structure of companies such as Anthropic and Mistral, it is clear that most investors are not the standard venture capitalists (such as Andreessen Horowitz) but also include “vendors”– large cloud providers (such as Microsoft and Amazon).

This creates a round-trip effect because a cloud provider invests $2 billion into an AI startup. Then the AI startup spends that $2 billion purchasing cloud credits to the same provider. The equity in the AI startup investment ecosystem not only represents capital but also secures supply chain relationships. Without a strategically coordinated relationship with a cloud provider, a startup may be at the back of the queue to buy the newest chips just because they come from that provider. This creates a complicated relationship when dealing with AI startup investor relations. Founders must manage financial investors who are looking for 10x returns and also manage strategic investors who want to ensure that their software runs on specific hardware or that their software integrates into the company’s office software.

Making sense of AI startup equity distribution for the team

Let’s discuss the individuals who are responsible for coding. In Malaysia and throughout Asia, it is extremely difficult to identify a PhD who has an understanding of large language models (LLMs). Due to their scarcity, these are the most in-demand employees in the tech sector and they also demand the highest salaries. For AI start-ups to attract LLM programmers, equity contributions to employee option pools must significantly in excess of the traditional ten to fifteen percent of the company. AI companies also need to constantly refresh or increase their employee option pools based upon the constant competition from both start-ups and big tech companies. Example likes Meta, who can offer salaries in the millions.

There is also a drive to dilute founder’s equity as the competition for talent continues to increase. Each founder must choose whether to hold on to as much equity as possible or to dilute their holdings in order to attract a researcher who could potentially double their production capabilities. Most will choose the second option. In the AI sector, holding 5% of a rocket ship is more advantageous than holding 50% of a paper airplane.

Navigating the AI startup funding rounds and the “Moat”

During an investment round regarding artificial intelligence start-up companies, the primary areas of concern for potential investors is the so-called “moat.” That is, “If Google were to take this technology to market tomorrow, would you be able to survive”? Traditionally, the answer has been one of two concepts—network effects or proprietary data. Today, more often than not, the conversation revolves around how the startup has created a “moat” through its ownership structure, and by strategically partnering with the right people.

AI startup ownership structure are evolving to support those companies that are able to leverage not only a superior algorithm. But also the operational “grit” needed to address their immense capital needs as well. Consequently, today’s considerations extend beyond just determining who is the smartest PhD to include which companies have the optimal AI startup corporate structure to sustain the hundreds of millions—and ultimately billions—of dollars that will flow through an AI company. When determining the viability of an AI startup, whether you are a founder in Kuala Lumpur or a researcher in Singapore, knowing the “cap table” will provide you more insight into the future of that company than will any pitch deck. The cap table provides a roadmap on how and where power resides.