Margin Points - Arnold Engel

Sam Altman announced last night at a YC gathering that OpenAI is offering all current YC companies a $2M investment offer in the form of OpenAI tokens. The $2M will convert in each startup’s next round of financing and is done in the form of a SAFE.

There is some very clever handling of adverse selection in here.

The tokens here can be spent quickly (by Demo Day) or slowly. Startups that are burning tokens quickly will prefer this deal and that is the type of company that OpenAI is most interested in backing and seeing grow quickly. Startups in the batch will explore different applications and approaches and the successful ones will spread what works. There doesn’t appear to be a timeline on accepting or rejecting the offer during the batch, so as early token monsters eat tokens and find success there can be new converts.

While there is certainly pivoting during the batch, for many they’ll already have shovel-ready places to burn these tokens. For others who are clear on their business model and on the fringes of product-market fit, it’s an opportunity to reflect on where this new input could help change the shape of what they can do for customers. Boom—OpenAI also gets free mindshare, even from those that don’t take the deal.

This offer is made easier for OpenAI because they have high gross margins on the API tokens they are using in this deal. If they are at a 70% gross margin, a $2M retail headline figure for value represents closer to $600K of actual cash costs to OpenAI.

There is an opportunity for startups to arbitrage the tokens to other companies and capture the cash and book the revenue today. That pushes out the timeline on which they need to raise their next round and see OpenAI convert onto their cap table.

While that might appear to be an issue for OpenAI—the startup is just brokering the token—that misses some of the value. The startup would be in a great position to build a thicker wrapper around the experience for the customers it was brokering the tokens to and would be developing or deepening customer relationships. It’s not without positive residual exhaust from the effort—this is a far cry from drop shipping things on Amazon.

Demo Day and the incentives for the startup that exist outside of this deal would push to maximize the value, rather than just dumb-piping the tokens to others in a modern revenue financing scheme. YC having screened these companies and working with them during regular office hours cuts a lot of the risk for OpenAI.

The offer will also cleave a line in the sand for the batch. The maxxers and the minners? Everyone will be judged at Demo Day with this X factor: “what did you do with your tokens?” or “oh, interesting you didn’t take them?”

Demo Day also provides a nice opportunity for OpenAI to continue distribution momentum. Everyone can see what was built and through all the launches on X as well, each providing a showcase for the ingenuity of how these resources were used. If these early signs are positive, OpenAI can do something like this with the next batch—it has bought the option with being the first mover on this to keep doing it.

Announcing it on a visit and casual interview on stage with a YC partner makes it seem casual and not deeply considered, which actually makes it much more powerful.

Of course, a different type of offer like this has been popular for a long time. Free credits to get startups started on Stripe, Zendesk,¹ AWS, Google Cloud, etc. have been a mainstay. a16z speedrun offers $7M worth of free credits to startups in the program. The trade is conceptually different in subtle but important ways. Free credits to get you started is a marketing concept that is proven. It’s free and you either want to use the product or you don’t. The scope of consideration is much narrower.

“I’m going to give you tokens and take equity on the bet that you’ll unlock disproportionate value from them” is something different to consider. In that consideration, it’s the startups who take the deal that are the most bullish on turning the tokens into company value. It’s aspirational, it’s forward-looking and it’s not something everyone can provide. “We’ll help change your valuation” is a different frame from “we’ll help save you money.”

It may open the door for more creative approaches and trades. Things like a studio-type fund, who takes equity and provides services for free (like an awesome launch video or branding) and pays its own employees with carry and cash raised from LPs. Other vendors might get creative with their offers.

There’s a lending-to-customers concept at play here, although with an equity component. That usually comes with some adverse selection risk, but here it’s washed out by a much bigger potential upside for both sides.

SpaceX’s IPO anticipation is taking all the oxygen out of the room.

From CNBC:

“Nobody wants to be caught in the SpaceX blast radius,” said Renos Savvides, head of equity capital markets at Neuberger Berman. “If you’re a smaller IPO and you’re on the road the same time as SpaceX, no one is going to pay any attention to your deal.”

Wait your turn, OpenAI & Anthropic. And if you aren’t part of the big three for the next year, well, your luck might be in the care of the great IPO sorting hat.

Obviously, there are ways around this if you are an aspiring public company. You can grow huge on your own and IPO or merge with a public company or merge with other private companies and IPO together.

Merging seems very underrated at this moment.

SpaceX is IPOing into being one of the top 10 largest public companies in the world, which is astounding. Saudi Aramco had a mega $25B IPO fundraise in 2019 and is also in the top 10, but that was a well-established business in a well-established market.² Nothing else comes close to SpaceX.

SpaceX itself is the product of the Musk bundle. Starlink, which accounts for the majority of its revenue today, and xAI and X are all part of it. The Musk bundle kicked the door open for others to also combine.

The first mega merger company will get attention, and disproportionately. A first-merger-advantage, if you will.

Modern merger theory was built on efficiencies and cost savings first. Then ‘synergies’ that would unlock greater revenue growth. Then ‘moat’ or defensibilities to make it harder to compete with and allow for higher prices or lower costs, translating to more profit.

But now, another thing? Attention. Specifically attention from investors. Big, moving targets draw attention. The new bundle is not for efficiency but for hyper-relevance and attention. Attention breathes oxygen into the room for companies.

AI-for-management-information cuts the downsides of traditional bloated conglomerates. Sure, there are some efficiency gains too, but that’s not the driver here. If your revenue is being valued at a high multiple, anything you can add to it will likely be valued more highly in your orbit than on its own.

Big public companies also don’t have much of the downside of smaller ones. Information and financials can be categorized into bigger buckets that obscure margins for specific product lines. AI-native businesses are also easier to integrate with and run through an M&A process.

It’s not that hard to find tie-ups that are beneficial in terms of market, inputs, or customers that can be packaged. We’re quite a ways away from seeing our first slop-merger, purely executed on the say so of a model. Executives have more time to think deeply and pursue courtships.

Focus? Well, each offering of sufficient size today already has the ‘mega-menu’ on their website listing every possible solution to anything related to their area and any possible potential customer that could possibly walk through the door. The era of focus is now one of vague sense: Stripe is money stuff, Salesforce is sales stuff, Adobe is design stuff, Uber is convenience stuff. Details get filled in after the mental shortlist for a buyer has been set.

Combinations can happen now to leapfrog and find oxygen in the public markets. At a different point in the cycle, they’ll be about finding a port in the storm.

Intrepid reporting from the WSJ tells us that hacky sacks are officially back in:

The little bean bags that once appeared relegated to the same cultural graveyard as mixtapes and Blockbuster have taken over high-school sports fields, hallways and sometimes classrooms seemingly overnight.

Cue nostalgia. Cue Backstreet’s back?

This story goes beyond nostalgia.

There is no brand associated with this, and that makes it feel more genuine. There is a Hacky Sack brand, but the brand name has become generic. Technically, it’s a foot bag—but that seems too odd a concept, like some sort of plastic wrapper you’d put on a walking boot to cover it in the shower.

Products from a slew of different makers are easily available on Amazon and Etsy and everywhere else hacky sacks are sold. No one is really pushing a brand in a major way, which is almost quaint. Retailers are happy to sell here and drive local awareness, but there probably isn’t gold sitting in the middle of hacky sacks.³

Contrast this no-brand scenario with others. Hoverboards were hot for a moment as well but tapered off quickly when some started catching fire. Without known brands behind them, the safety risks and financial risks of buying a dud started to outweigh the cool factor. Had a known electronics brand stepped in earlier or an upstart leaned heavily into establishing a trusted brand, the story may have turned out differently.

We are somewhere on that journey now with peptides. Brands for distribution and manufacturing are starting to coalesce in the peptide space. They will likely drive for differentiating if at all possible to protect their own investments in the space and prevent unbranded generics from cutting corners that could lead to customers being turned off from the category.

Both peptides and hoverboards have higher costs and real safety issues. Hacky sacks are gleefully cheap, and any safety issues are entirely self-inflicted. They are able to stay in a brand-less state indefinitely without damaging the category or impeding the growth prospects.

This isn’t even the product of nostalgia marketing, like Coinbase’s Backstreet Boys Super Bowl ad or Tin Can’s appeal to millennial and Gen-X parents. It’s driven by the kids, not the parents.

In all of it, a generation is getting exposure to some business thinking: it’s nice to have both a single-player mode and a potential group option when you play with others on the playground. Come for the tool, stay for the network.

---

Like these essays? I write a trio of essays on wide ranging topics delivered four times a week. Subscribe and forward it to your friends. Have feedback, thoughts to share or comments? Comment or reply.