AIOZ: what's it worth?
(nothing in particular)
“a decentralized peer-to-peer CDN”
Of all the content you consume on the internet, streaming video is one of the most expensive to deliver.1 Startups that try to take on YouTube tend to find themselves bleeding out from the delivery costs well before achieving any sort of widespread adoption. AIOZ promises to solve this by building a peer-to-peer worker network of previously underutilized storage/compute/bandwidth, coordinated on their own layer 1 blockchain, all paid for with the eponymous token. By avoiding upfront capital expenditure and running on equipment available for a low marginal rate, they hope to undercut the existing (centralized) solutions.
Hey it’s a decent premise. I mean, BitTorrent works okay, and no one even pays for that.
Sadly for investors, it’s absolutely impossible to value the fucking token.
Negative 8.26% APY forever
AIOZ ICOed at the beginning of April. In a stroke of terrible judgement, they used the opportunity to promote the token-earning potential of their streaming site and worker node—both currently running on the testnet, earning testnet tokens. Cue a week-long monsoon of confused users flooding into the official Telegram channel asking how they could exchange their fool's gold for the real thing. It’s an oversight that’s not really material to our goal, but it sets the tone nicely.
There are three types of work performed by the network: storage, transcoding, and delivery. Somehow they've come up with proof-of-work mechanisms for each of these; technical details aren’t the focus today so we’ll just trust them on this.
Their blockchain (based on Tendermint) uses proof-of-stake to elect 21 validator nodes, who are compensated by mining new tokens equivalent to 9% annual inflation. Why inflation instead of transaction fees or revenue share? Well, how else are you supposed to incentivize staking before there are any transactions or revenue?2 Naively, we might assume all AIOZ not currently being used for payment will tend to be staked to avoid inflation, shrinking the money supply.
They haven’t really published anything else we can use to constrain the token valuation. No details on pricing mechanisms for the work done (I guess you could try to reverse engineer it from the testnet). Nothing directly tying token supply to the amount of work done by the network. Is it possible for an investor to come up with a target price for the token grounded in something other than speculation? So far, we know it has 9% inflation, and of course the other forces acting on the price are the inevitable speculation and velocity of money. Depending on the work pricing mechanism, there may be some token price inertia from the workers as well.
Maybe math will help
We can try to estimate the proportion of tokens that will be staked at any time. Plotting the real APY of staking (corrected for inflation) in terms of the proportion staked, a target APY of 5% (competitive with historical equities returns) results in a little over 60% staked.
Great. So tokens worth x will probably pay out around 0.05x per year, but what the hell is x? I'm not smart enough to write out all the relevant equations and take partial derivatives to determine pricing dynamics, so let’s try a toy model and walk through some scenarios.
What if
In our toy model there are two participants: a worker who owns tokens that they have just received for work, and a customer who wants to pay for some video delivery. There is also a recent price for tokens and a recent fee for work (in terms of tokens). The worker needs to sell some tokens to cover their cost of operations, but they are presumably making a profit and don't need to sell all of them. What they do with their profit depends on where it is more useful. The one thing they will not do is hold tokens unstaked indefinitely, because of inflation. If staking offers better returns than those available outside of AIOZ, they will stake, otherwise they'll withdraw funds. As above, below a certain effective APY all earned tokens are quickly sold, ensuring that a relatively fixed proportion of tokens remain unstaked and in rapid circulation.
What happens when demand from the customer varies over time? The worker wants to sell a different quantity of tokens than the customer wants to buy, so the price fluctuates. Either party might try to hold some unstaked tokens over time to profit from price swings by buying low or selling high, but again the inflation cuts into those profits, demanding a wider bid/ask spread to be worthwhile, and limiting the dampening effect.
What happens when a speculator enters the market? For the most part, the worker and the customer don't notice. They are already holding for short periods of time relative to flow because of inflation. When there are inflows, the worker gets a mild subsidy, and when there are outflows the customer gets a mild discount.
You can work through it yourself and see that the mechanisms in our toy model hold even as you add more participants. But still there’s nothing to concretely tie the price of the token to the value of work the network provides.
It’s broken
Where does this leave AIOZ holders? Without any fundamental value, they’re really at the mercy of speculation. In the best case, speculation drives the price up nice and high, and then actual demand picks up. Increasing demand does create positive price pressure, so some holders can sell their tokens to customers without crashing the price. If there's enough payment flow, and they trade out slowly enough, it could go okay, with speculators taking a small cut of profits that would otherwise go to workers.
In the much more likely outcome, some speculators get impatient or spooked and pull funds out faster than payment picks up. The price crashes. The actual users of the network happily continue at the new low, and holders are stuck until a bigger sucker comes along. All the while inflation is chugging along, trying to drive the token price just a little bit lower.
Fix it
If you want to give the token some fundamentals, just make it behave like a normal equity.3 Pay stakers a share of worker fees. Any valuation is then based on an estimation of future network revenue. Obviously, the proportion can’t go too high without affecting the whole network’s competitiveness, so it probably makes sense to have a governance mechanism to tune the proportion. If you don’t like that one there’s a multitude of other ways to patch things up (and I’m sure the comments will be full of them).
Or don’t
On the other hand, maybe you think nothing is broken here. Yes, the token price is totally independent of the value of the network, but that actually makes it easier to arbitrarily drive the price up. Crypto runs on hype, and a high market cap is a great way to get attention. If you’re a fundamentals investor, what the hell are you doing buying shitcoins anyway? This world is for people who want to get rich quick or be margin-called trying.
Whatever your feelings are, as things stand it’s impossible to invest in AIOZ—all you can do is gamble.
An estimate from Amazon for live streaming to 5000 viewers puts the cost at $757 per hour.
LivePeer has a similar mechanism, but with the refinement that once 50% of tokens are staked inflation goes to zero.
Although you risk annoying governments who have lots of rules about equity offerings.