OpenAI, the company behind ChatGPT, expects to end 2025 with an annualized run rate of revenue over $20 billion; its fourth quarter revenue will be about $5B. OpenAI’s CEO Sam Altman predicts revenue will grow to hundreds of billions by 2030. 2030 is also when the company is guiding positive free cash flow. Essentially, the company will have cash after paying all operating and maintenance costs. The common definition of free cash flow is Cash from Operations (CFO) minus Capital Expenditures.
Based on these growth projections, OpenAI has $1.4 trillion committed to data center infrastructure projects over the next eight years (i.e., its capital expenditures). Altman was recently questioned about these staggeringly large commitments on Brad Gerstner’s BG2 podcast and became defensive. Altman pointed to multiple avenues for revenue growth, including:
Altman became frustrated when pressed, stating, “If you want to sell your shares, I’ll find you a buyer. I just......enough.” Instead of relying on market analysis and financial projections, Altman is implying that their business model must be good because investors are buying their stock. As the CEO of a company with negative free cash flow, Altman needs investors to remain excited about OpenAI’s future. This allows OpenAI to continue to go to the debt and equity markets for funds (cash) to make up for the negative free cash flow.
How realistic are OpenAI’s growth projections? A natural comparison for OpenAI is the early years of Alphabet (GOOGL), the parent company of Google. Google search looks to organize the web and provide answers to your queries. In some ways, ChatGPT is the next evolution of search, as it will also answer your queries based on information found online. The answers are more thorough and conversational than traditional search, but in some ways, the end game is the same. ChatGPT, and other Large Language Models with which it competes, like Google’s Gemini and Anthropic’s Claude, have many applications beyond traditional search, e.g., scientific research, but the comparison is justified.
ChatGPT became available to the public in November 2022, meaning 2024 was T+2. According to multiple industry reports, OpenAI’s estimated 2024 revenue was $3.7B. Google Search went live in September 1998, making T+2 the year 2000. In 2000, GOOGL earned $19.1 million in revenue. In 2001, revenue grew more than 350% to $86.4M. 2025 is the parallel year for OpenAI, and applying the same growth rate results in $16.7B in revenue. As referenced above, OpenAI is projected to end 2025 with a run rate of $20B, so $16.7B for the year is a fair estimate. We can continue these projections by applying GOOGL’s growth through 2008 to estimate ChatGPT’s revenue over the next 8 years.
To further the comparison, we can use GOOGL’s CFO margin as a proxy for OpenAI’s margin to estimate the company’s CFO margin. OpenAI needs this CFO to pay for the $1.4T in infrastructure commitments. Using this simple and speculative model, we have OpenAI generating a total of $1.5T in CFO over the next five years, which is just more than its commitments over the next eight years.
If OpenAI somehow grows in a similar pattern to GOOGL, one of the most successful companies of this century, while competing with GOOGL and others, it will generate the cash to meet the $1.4T in commitments. Of course, the required 190% annualized revenue growth means revenue would reach a ridiculous $2T by 2030, or just under 2% of global gross domestic product. For Comparison, Alphabet’s estimated 2030 revenue is about $600B.
Two trillion in revenue is mind-blowing. So, what kind of growth does OpenAI need to generate $1.4T in CFO over the next 8 years? Altman is predicting hundreds of billions in revenue by 2030, implying less than $1T. If we grow OpenAI’s revenue 100% per year for the next eight years, we project $535B in revenue by 2030 and $2T by 2032. This generates the $1.4T in CFO needed but relies on historic growth and on total revenue matching or exceeding those of the largest companies in the world.
Alphabet was able to achieve this growth in the early 2000s. Google search quickly displaced other search engines, became a verb (“Google it”), and drove advertising dollars away from legacy media such as newspapers, radio, and television. Google was able to do this by delivering cleaner, more relevant search results, while providing a better return on investment for advertisers. The website’s ability to collect data on its search results and provide an easy way to purchase search terms helped capture share of the $300B advertising market (estimated size in the early 2000s). Google acquired YouTube in October 2006 and released the Android operating system in September 2008, so this growth was largely due to desktop search.
Google’s business model was simple at the time: to take advertising share from legacy platforms by providing a better alternative. Altman’s suggested avenues for revenue growth show that OpenAI’s model is not so simple. ChatGPT's growth is likely, but OpenAI is competing with a multitude of players, including Google’s Gemini, which are unlikely to be displaced as easily as late-1990s start-up search engines like AltaVista and Excite. OpenAI has invested in clusters of AI chips to build out high-powered data centers and related infrastructure. Leasing out this computing power makes sense and is already done by Alphabet (Google Cloud), Amazon (AWS), Microsoft (Azure), and others. OpenAI’s data centers are native to AI, i.e., built to run AI models, and may therefore be more efficient than legacy public clouds. However, the competition will be fierce. It’s clear that following Google’s footsteps is far from guaranteed.
There is an arms race to build out the most powerful and efficient computing to power artificial intelligence. This means that instead of fearing excess capacity and showing restraint, companies are likely to overbuild for fear of falling behind. Meta came under this very scrutiny after revealing its intentions to spend over $100B in capital expenditures next year. OpenAI is not exempt from scrutiny just because it’s not a public company, meaning its commitments to spending are not guaranteed. If OpenAI falls short of its $1.4T commitments, it will ripple through the market, particularly companies that are depending on OpenAI infrastructure spending for growth. The underlying growth drivers of a strong investment should never rely on one company’s commitments, no matter how promising the technology.
Source of data: Bloomberg
For informational and educational purposes only.
Carnegie Investment Counsel (“Carnegie”) is a registered investment adviser under the Investment Advisers Act of 1940. Registration as an investment adviser does not imply a certain level of skill or training. For a more detailed discussion about Carnegie’s investment advisory services and fees, please view our Form ADV and Form CRS by visiting: https://adviserinfo.sec.gov/firm/summary/150488