#316: Terra Blowup Sends A Warning To Crypto Markets, & More

Crypto markets deteriorated sharply last week as the largest algorithmic stablecoin, UST, broke its peg to the dollar, triggering a death spiral that eventually led to its collapse. Prior to the collapse, which began last weekend, UST and Luna, its companion cryptoasset that is intended to be used to maintain the price stability of UST, were valued at approximately $18.7 billion and $28 billion, respectively. As of this writing, their combined market cap is roughly $2 billion, down 96% over the last week. Even the $2 billion valuation appears spurious because several exchanges either have stopped trading or have delisted the assets, and validators have at times stopped or limited transactions on the underlying Terra blockchain.

In our view, Terra’s unraveling stems from its price stability mechanism. Unlike USDC, a fully collateralized stablecoin backed 1-to-1 by cash and cash-like instruments in a bank, Terra sought to maintain UST’s price peg to the dollar algorithmically using Luna to absorb changes in demand for UST. Although that strategy worked in relatively stable market conditions, a broad-based selloff decreased the demand for all cryptoassets, including UST and Luna, enough so that the arbitrage necessary to stabilize the UST price became unattractive to investors. The situation worsened as the Luna Foundation Guard, an entity established to defend UST’s peg in such a scenario, deployed the $3 billion BTC in its war-chest to backstop the peg, and failed.

In addition to causing the crash in UST and Luna, Terra might be the largest layer 1 blockchain failure in crypto history and has impacted every protocol and cryptoasset associated with it. Most of the approximately $28 billion in value locked in DeFi on the blockchain as of May 7th has evaporated. Built upon a proof-of-stake blockchain, Luna’s price crash also has introduced negative externalities for the security of the network, forcing network operators to take the blockchain offline to protect any remaining on-chain value.

Terra’s demise is one of the largest fiascos in crypto market history as measured by market capitalization affected relative to total crypto market cap. Compared to the Mt. Gox hack in 2014 that stole 7% of outstanding bitcoin, Terra’s collapse has destroyed roughly 3% of crypto’s total market capitalization.

As the Terra blowup prompts more stringent regulation of crypto in general and stablecoins in particular, we believe that more secure and conservative blockchains like Bitcoin and Ethereum will continue to gain market share in the crypto market ecosystem.

Sam Bankman-Fried, Founder and CEO of crypto currency exchange FTX, has taken a 7.6% position in Robinhood according to a 13D filing this week. While the filing notes Bankman-Fried does “not currently have any intention of taking any action toward changing or influencing the control” of Robinhood, 13D filings typically are associated with activist investors, unlike 13G filings, which flag passive holdings. Bankman-Fried’s $648 million investment gives him a vote of 2.8%, well below the Founders’ > 60% ownership that should guarantee the company’s independence, according to the Wall Street Journal.

Earlier this year, FTX announced plans to offer equity trading on its platform. This week, FTX President Brett Harrison followed up with a screen shot showing a trade of Apple stock on the FTX platform. Bankman-Fried’s move seems to be a strong signal of support for Robinhood now that its share price has dropped 70% from its IPO last July.

FTX and Coinbase seem to be parting company in their strategy for equities. Asked about the “strategic advantage of owning a traditional securities platform” during its earnings call last week, Coinbase responded that it does not “plan to offer traditional securities” unless it helps to “massively accelerate crypto adoption”.

DeepMind recently published a paper on Gato, a multi-modal, multi-task AI model that can perform 604 tasks, from playing Atari to stacking blocks with robot arms. The “foundation model” is trained uniquely on a variety of control, vision, and language datasets and can perform a wide range of downstream tasks with the same neural network. With just 1.2 billion parameters, the model scored higher than the 50% expert score on 450 of 604 tasks. In contrast, the number of parameters in OpenAI’s GPT-3 is more than 100 times higher at 175 billion. Moreover, as it scales, DeepMind is likely to make Gato even smarter.

Informed by Stanford’s analysis of foundation models, ARK’s Big Ideas 2022 highlighted their financial potential with an estimate that the AI economy could generate ~$25 trillion in enterprise value by 2030.