OpenAI – Unpalatable Confection.

OpenAI wants to have its cake and eat it.   

• What was going to become a for-profit entity subject to all the forces that hone companies into efficient cash printing machines that benefit their shareholders and employees will now be a Public Benefit Corporation (PBC) that will give OpenAI extra space to shield management from the consequences of its bad decisions.
• This is yet another sign that investors are so desperate to invest in AI that they are willing to overlook shortcomings that, at some point in the future, could easily turn the value of their investment into a tiny fraction of what it is today.
• Instead of becoming a proper for-profit corporation, OpenAI is now pursuing a rare corporate structure known as a Public Benefit Corporation which on top of the usual fiduciary duty to shareholders and debt holders, the company must also demonstrate that its activities also benefit the public and society more widely.
• This structure is pretty rare but one I have heard of is Black Rifle Coffee which is a coffee roasting company that also supports US veterans, law enforcement and first responders as its public benefit.
• OpenAI’s, of course, will be its current statement which is to “ensure that artificial general intelligence benefits all of humanity” but unlike Black Rifle Coffee, many of OpenAI’s actions are in direct contradiction to this statement.
• The best examples are its increasing secrecy (see here) and its actions in trying to stop users from figuring out how its technology works (see here).
• In my opinion, compliance with its statement would involve the open-source release of GPT-4 and its derivatives, disclosure of the data set that was used and the techniques that have been developed to ensure that the models work better with fewer mistakes.
• Instead, the public received no data and threats of disconnection for anyone with the temerity to try and find out how the technology works.
• Consequently, how OpenAI benefits the public, society and humanity more generally remains a complete mystery
• A PBC also has more onerous reporting requirements because it has to demonstrate that it is fulfilling its public benefit alongside the normal financial disclosures that apply to shareholders.
• However, these dual fiduciary duties can easily come into conflict with one another which was pretty much why OpenAI almost imploded a year ago.
• Furthermore, the public benefit can also give management places to hide from hostile takeovers, shareholder activism and other external pressures that can disrupt the normal functioning of a company.
• In my opinion, these disruptions almost always occur because the current management of the company is performing badly and the events force management to perform better or sell the company to someone who can fix it.
• OpenAI will be able to underperform financially and operationally and still refuse to change using its PBC mission as the reason for its underperformance.
• If OpenAI does go down this route and does not open source its models, datasets and techniques, then in my opinion it will be in breach of its public benefit mission and will have only adopted this structure in order to avoid the consequences of management’s bad decisions.
• This is where Meta has an advantage as it has already outsourced its largest and best model and in this regard is doing more to bring advanced AI to all of humanity even though it is a fully for-profit company controlled by its founder.
• By contrast, OpenAI seems to want to have all of the protections of being a PBC but not fulfil any of the obligations.
• In my opinion, this sets the company up for more problems, but it does represent some reduction in the corporate governance risk that almost killed the company one year ago.
• This is because the non-profit part of the company will cease to govern the whole company but will instead be a side activity which uses the technology to fulfil the PBC obligation.
• However, there is still plenty of space for conflict because to do its job, the non-profit part will need to do all of the things that the for-profit piece does not want.
• Hence, there is still plenty of scope for problems and I do not think that a PBC structure represents a good way out of the current conflicts of interest although there is some reduction in corporate governance risk.
• OpenAI is currently valued at $157bn putting the company on 30x 2025 revenues in an optimistic scenario.
• This assumes that none of the above risks materialise and that revenues grow very quickly and fat profits follow.
• With the brains of the operation gone and possibly gearing up to compete with OpenAI directly and rapid growth in competing offerings, this proposition is becoming more fanciful with every passing event.
• OpenAI is no longer miles ahead of the competition and with Meta’s party-blowing release of its flagship model and its weights into the open-source community, the stage is set for rapid price erosion.
• This is why I remain nervous about the valuations being paid for these companies and am staying away from them.
• If I were forced to invest in this area, it would be Nvidia which actually has revenues and profits now or Qualcomm (already owned) which is in an excellent position to benefit as generative AI starts to be implemented at the edge.