OpenAIs structural transformation and valuation

From: allin

OpenAI is reportedly undergoing a significant structural transformation, converting into a for-profit public benefit corporation (B Corp) from its original non-profit structure. This change has drawn considerable attention, particularly regarding its valuation and the compensation of its key figures like Sam Altman.

Proposed Structural Change

OpenAI is in the process of converting into a for-profit benefit corporation (B Corp) [06:54:33]. A B Corp is a variant of a C Corporation where the board of directors is required to act as a fiduciary for both shareholders and the company’s stated mission, which can include external stakeholders like the environment or society [07:07:34]. This structure is intended to signal to investors, the market, and employees that the company cares about more than just profit [07:37:34].

Reuters has cited anonymous sources indicating that the plan is still being developed with lawyers and shareholders, and the timeline remains uncertain [07:56:39]. The discussed structure suggests that the non-profit entity will continue to exist as a minority shareholder in the new company [08:08:44].

A significant change being discussed is the removal of the 100x profit cap for investors [08:25:01]. This means early investors, such as Vinod Khosla and Reid Hoffman, could see their 100x returns turn into 1,000x or more [08:33:02].

Valuation and Financials

OpenAI is reportedly valued as high as $150billion<aclass="yt-timestamp"data-t="08:47:06">[08:47:06]</a>.Thisvaluationhasbeenspeculatedformonths,withreportsindicatingafundraisingroundintherangeof$6 to $7 billion [03:40:15].

The company has shown rapid revenue growth:

• $28 million in revenue in 2022 [10:54:19].
• $1.3 billion run rate in October 2023 [10:49:12].
• $2 billion in 2023 [10:45:10].
• Projected to be on a $3.4 billion run rate as of June 2024 [10:38:15].
  • Whisper numbers suggest a potential run rate of $4billionto$6 billion for 2024 [11:11:08].

Based on the $3.4billionrunrate,the$150 billion valuation represents approximately 50 times the current top-line revenue [11:02:18]. If the run rate were $5 billion, it would be around 30 times, and if it’s growing at 100% year-over-year, it could be 15 times next year’s revenue [11:23:44]. It is suggested that 15 times forward Annual Recurring Revenue (ARR) is not a high valuation for a company with such strategic opportunity [11:38:51].

Sam Altman is reportedly set to receive equity of 7%, which, if the $150billionvaluationistrue,wouldplacehisstakeataround$10.5 billion [08:38:57].

Arguments for the Valuation (Bull Case)

The bull case for OpenAI’s $150 billion valuation centers on its strong market position and ability to extend its technological lead:

• Moat in Model Performance and Infrastructure: Significant capital raised will be aggressively and strategically deployed to improve model performance and extend their advantage [09:08:29].
• Leadership in AI: OpenAI is considered the leader in the field, possessing the most advanced AI models, the best developer ecosystem, and superior APIs [11:54:16].
• Product Innovation: Products like the 01 model, voice application, and Sora (though not publicly released) are considered ahead of the pack [09:43:01].
• Multi-Trillion Dollar Market: The AI market is projected to be multi-trillion dollar, with opportunities across numerous verticals, applications, and products, allowing OpenAI to become a global player [10:10:48].
• Advancements in AI Capabilities:
  • Chain of Thought: The 01 model uses a “Chain of Thought” approach where the model asks itself how to answer a question, then breaks it down into steps, calling the LLM to fill in the blanks and link them for a comprehensive answer [18:18:49]. This is seen as a game-changer and a new paradigm for AI systems [19:13:90].
  • Agents: This Chain of Thought will lead to AI agents that can break down objectives into tasks and execute them, similar to complex knowledge work done by humans in operations centers [20:17:51]. OpenAI has indicated “PhD-level reasoning” is next on their roadmap, with agents not far behind [20:34:03].
  • Impact on Knowledge Work: AI tools like the 01 model can perform analysis tasks that previously took human analysts days, within minutes [22:21:49]. This is seen as completely revolutionary for ad hoc and repetitive structured knowledge work [22:57:42].
  • Disruption of Systems of Record (SoR): It’s argued that traditional systems of record (like Salesforce, Oracle GL, Workday Financials) may become obsolete, replaced by direct data processing and AI agents that can handle complex workflows and generate outcomes without fixed processes [25:48:47].

Arguments Against the Valuation (Bear Case)

The bear case for OpenAI’s valuation highlights potential weaknesses and competitive threats:

• Commoditization of Technology: The underlying frameworks for models are well-described and available in open source, with viable open-source models (like Meta’s Llama 3.2) potentially being as good or better than OpenAI’s at any given time [13:07:07]. This could drive the economic value of models towards zero, creating consumer surplus but making monetization difficult [13:41:20].
• Front Door Competition: Major players like Meta and Google could become more aggressive in integrating their AI (Meta AI, Google’s version in front of search) into their widely used applications (WhatsApp, Instagram, Messenger, Facebook, Threads), reducing the need for users to go to a separate ChatGPT-like app [13:57:33].
• Data Scarcity: Models may eventually run out of viable data to differentiate themselves, leading to a race around synthetic data [15:03:07]. This becomes a cost problem, where large companies with effectively infinite money (Facebook, Microsoft, Amazon, Google, Apple) have a significant advantage over startups [15:14:52].
• Human Capital Turnover: Significant high-level churn at a company supposedly on a “rocket ship” trajectory, coupled with high liquidity, raises questions about why technical talent would leave if they were enamored with what they were building [15:31:08]. The departure of individuals like Ilya Sutskever, Jan Leike, Sean Sherman, and Mira Murati is noted [00:34:03]. Greg Brockman is also on extended leave [01:08:00].

Summary of Bear Case

The bear case rests on:

• Open-source competition [16:17:58]
• “Front door” competition from major tech companies [16:20:00]
• The shift to synthetic data [16:20:00]
• Executive turnover [16:23:44]

Controversy and Implications of the Conversion

The structural conversion, particularly the shift from a non-profit to a for-profit entity, has raised eyebrows.

Chamath Palihapitiya points out that if this model “passes muster,” it could set a precedent:

• A company could start as a non-profit, paying no income tax, allowing it to outspend and outcompete rivals [42:46:17].
• Once it achieves dominance, it could convert to a for-profit corporation [43:17:09].
• The reverse could also be true: a for-profit company could convert to a non-profit to gain economic advantage over competitors, then flip back when they wither [43:32:04].

This “hack on the tax code” could lead to many companies adopting similar strategies if deemed legal [43:21:40]. The Mozilla Foundation is cited as a prior example of this conversion [45:58:39].

Ethical Concerns

The situation is contrasted with Elon Musk’s $55billionpaymentpackageatTeslabeingtakenawaybythejusticesystem,whileSamAltmanstandstomake$10 billion from a structure “definitely pushing the boundaries” [44:50:39]. This disparity is seen as “not how the system should probably work” [45:06:55].

Elon Musk’s Role and Compensation

Elon Musk, a co-founder of OpenAI, invested $50 million as a seed investor in the non-profit entity [46:05:07]. Despite the company’s re-factoring and Sam Altman receiving a substantial compensation package, Elon Musk is not receiving any equity in the new for-profit structure [46:50:04].

Sacks argues that if OpenAI is cleaning up its organizational chart and giving Sam Altman a CEO option grant, it should also provide “fair compensation” to Elon Musk for his foundational investment [46:51:39]. It’s perceived that OpenAI is effectively telling Elon Musk, “if you don’t like it, just sue us” [47:20:30].

A counter-argument suggests that Elon Musk was given the opportunity to invest in the for-profit side alongside other investors like Reed Hoffman and Vinod Khosla but declined to participate [48:31:07]. However, it is argued that Elon shouldn’t have been obligated to put in more money; his original $50 million investment was made with the expectation of a philanthropic purpose, which is now being changed [49:05:09]. Therefore, if the purpose changes, the company should “make things right with him” [49:21:05].