Arthur Hayes’ New Article: The Fed’s Triple-Kill Setup – The More Powell Struggles, the Crazier the Crypto Bull Run
Author: Arthur Hayes
Compiled by: TechFlow
TechFlow Summary: In the evolutionary history of crypto finance, every leap in asset paradigms has been accompanied by the birth of new underlying derivative components. In 2016, Arthur Hayes and his BitMEX team invented the Perpetual Swap, reshaping the global crypto derivatives trading landscape overnight; in 2023, he proposed in his essay "Dust on Crust" the concept of synthetic dollars constructed via spot longs hedged against inverse contracts, directly giving rise to the current synthetic dollar protocol Ethena (USDe), which has surpassed $10 billion in scale. Today, Arthur Hayes returns with his third grand vision: Flop Network.
Facing the pain points of geometrically exploding compute demand in the AI era, alongside the fragmentation, opacity, and lack of a unified spot benchmark in the traditional cloud computing market, Hayes proposes anchoring to physical Floating Point Operations Per Second (FLOPS). Through a decentralized inference verification network, he aims to establish a globally standardized spot exchange rate for compute, which will serve as the underlying base asset to build deliverable Compute Forward Units (FCUs) and cash-settled quanto futures and perpetual swaps.
This article thoroughly deduces the complete mathematical framework and market-making hedging logic for compute assetization and financialization, revealing the ultimate blueprint that crypto-native finance holds for reconstructing macro liquidity and compute pricing power in the age of AI agents.
I want to achieve a "Three for Three". I'm not talking about basketball; rather, I mean my ability to conceptualize entirely new financial derivatives or application scenarios and witness them evolve into financial industry standards.
My Historical Track Record:
Perpetual Swap: A Resounding Success (in the words of Poulat)
In May 2016, my team at BitMEX and I invented the Perpetual Swap (commonly known as “perps”). Perps quickly ascended to become the most traded financial derivative in the history of the crypto industry. Due to their massive liquidity and popularity, innovation platforms like Hyperliquid now drive 24/7 RWA (Real World Asset) derivatives trading precisely through this mechanism. [1]
Synthetic Dollar Stablecoin: A Resounding Success (in the words of Poulat)
In March 2023, I wrote an in-depth essay titled "Dust on Crust," systematically explaining how to construct a synthetic dollar through basis hedging using a "spot Bitcoin long + short perpetual contract." This synthetic dollar can be packaged into a stablecoin and widely used for payments and settlements throughout the crypto capital markets. The Ethena team adopted my concept and executed flawless engineering implementation. Ethena’s launched synthetic dollar, USDe, became the fastest-growing stablecoin in history, skyrocketing from $0 to over $15 billion in circulating supply in less than 18 months.
Flop Network Spot Compute Market: To Be Determined (TBD)
This summer, I first publicly unveiled my concept for Flop Network. $FLOP is the "food" for your AI agents. Agents will use $FLOP in commercial interactions because $FLOP can be directly exchanged for physical compute, specifically defined as Floating Point Operations Per Second (FLOPS). If Flop becomes the native payment and clearing network between AI agents, then logically, the exchange rate between $FLOP and compute will become the baseline anchor for global compute spot prices.
A truly globally liquid spot compute market will inevitably give birth to a derivatives market atop this spot benchmark. May this article serve as an authoritative blueprint for all ambitious entrepreneurs, guiding them to build compute derivatives markets that meet the demands of the new AI agent era.
Core Pain Points
There is currently no unified compute pricing benchmark worldwide. Although various indices exist on the market attempting to track the sales and lease prices of specific CPU, GPU, and TPU chips, the data samples are severely incomplete. Furthermore, many indices try to track the invocation prices of AI tokens (Token) model by model. However, chip prices and model token invocation prices are, in essence, merely secondary derivatives of compute. If I can accurately grasp the baseline price of compute (FLOPS), I can derive downward the intrinsic value of any chip and the reasonable cost of any model token. We should not waste energy collecting fragmented and imperfect chip and token pricing information; instead, we must focus exclusively on establishing a unified price discovery mechanism for underlying physical compute.
Solution: Flop Network Compute Market
Flop Network produces one block per second. Each block contains verifiable receipt proofs confirming that miners successfully completed an inference task for an agent. For every inference task, we obtain three core parameters used to precisely calculate the exchange ratio between $FLOP and physical compute.
The block data object includes:
Latency (in seconds): The actual time taken to complete the inference request
Completed Workload: The total number of floating-point operations actually executed (FLOPS)
Transaction Fee: The amount of $FLOP earned by the miner for executing the task
$FLOP / (Compute / Second) = [3] / [2] / [1]
Spot prices are only genuine and effective market prices when everyone is free to trade. On the supply side, as long as you own compute chips (whether GPU, CPU, or TPU), you can mine $FLOP by fulfilling inference requests for agents; on the demand side, as long as your wallet holds $FLOP, you can submit inference tasks at any time. Flop Network is an open-source network accessible to all of humanity. Therefore, the spot compute price formed here will become the globally recognized compute benchmark anchor.
Flop Compute Unit (FCU)
Before delving into cash-settled financial derivatives, let us first continue along the path of the "physical delivery" compute market. One core feature of Flop Network is allowing demand-side participants to pre-reserve compute delivery from miners for up to 30 days in advance. To enhance liquidity and tradability, the network standardizes compute into "Flop Compute Units" (FCU).
Imagine you know you have a large computational task scheduled for 7 days from now and wish to lock in the compute cost today. You can place a bid to reserve a specific amount of compute for delivery in 7 days. Once the 7-day period arrives, miners are granted a 24-hour fulfillment window to deliver the actual compute. Once your bid is matched and accepted by a miner, the reservation generates a standardized 7-day FCU object, which is functionally equivalent to a deliverable forward contract. If the spot compute price surges dramatically before fulfillment and you no longer need this compute, you can simply resell the FCU directly to other participants on the native Flop Network.
Below are concrete business examples of FCU operation in practice. Please stay tuned for updates to the official Flop documentation, where all technical details will be fully presented.
Specifications for a Single FCU Underlying Asset Object:
Latency: 1 second
Compute Volume: 1 PetaFLOP (quadrillion floating-point operations)
AI Model: Specified by the buyer
Delivery Time: 12:00 UTC, 7 days after creation
Note: FCUs only cover Inference scenarios and are not applicable to Model Training. The aggregate value of training compute clusters far exceeds the simple sum of individual hardware components; whereas inference tasks possess inherent high divisibility, which is the fundamental premise enabling the existence of a single benchmark unit like the FCU.
For each AI model, there can be up to 30 tradable FCU tenors on the market, corresponding to a maturity date each day. If you need to reserve 1 ExaFLOP (quintillion operations) of compute, you simply purchase 1,000 FCUs. As the buyer, to attract miners to take your order, you must publish a $FLOP bid you are willing to pay and make a full upfront payment for network verification; as the seller (compute provider), you publish a $FLOP ask, while simultaneously staking equity-weighted $FLOP as performance collateral to ensure that if you default, other miners can profitably step in on the spot market to fulfill the delivery. This collateral amount is dynamically determined by the network based on the total recent inference volume processed across the entire network, incorporating both the baseline $FLOP cost and a penalty multiplier to deter malicious fraud and delivery defaults. This mechanism creates a fully funded two-sided order book that completely eliminates spam orders: funded buy/sell quotes are finally settled on-chain, while the high-frequency continuous two-sided order book and matching engine operate off-chain, mirroring the operational mechanics of traditional spot markets.
The forward market inherently carries Failure to Deliver risk. As previously mentioned, to mitigate this risk, FCU compute providers must post enormous $FLOP performance collateral far exceeding the received inference fees. If a miner fails to deliver compute within 24 hours after maturity, their entire staked $FLOP collateral will be completely confiscated. Backup miners will then bid to step in and fulfill the order; they are willing to do so because the fee per PetaFLOP gained from taking the order is significantly higher than the prevailing spot price at that time. Should spot prices surge above the collateral amount such that no miners are willing to fulfill, the system will declare a default occurred. In this scenario, the buyer receives not only a full refund but also additional compensation equal to the entire confiscated collateral.
FRCR Futures Contract
FRCR: Flop Reference Compute Rate
Hourly-volume-weighted $FLOP / GigaFLOPS settlement rate
In the sections above, I deconstructed how real demand-side participants procure physical compute via $FLOP on spot and forward markets. Now let us turn to discussing how to construct cash-settled financial derivatives capable of being listed and traded on traditional finance (TradFi) and cryptocurrency exchanges. I will outline a complete contract design scheme and detail how market makers hedge long or short futures exposures.
Contract Specifications:
Contract Name: FRCR Quanto Quarterly Futures Contract; Delivery Maturity: 12:00 UTC on the last Friday of March, June, September, and December; Underlying Index: FRCR.
Contract Multiplier: $1 per point
Example: Assume FRCR Index is at 100 points
Contract Value: Multiplier × FRCR = $1 × 100 = $100
Currency of Margin: USD
If you go long 1 contract at an entry price of 100 $FLOP and settle delivery at 200 $FLOP, your PNL is as follows:
Multiplier × Number of Contracts × (Delivery Price - Entry Price) = $1 × 1 × (200 - 100) = $100
If you go short 1 contract at an entry price of 200 $FLOP and settle delivery at 100 $FLOP, your PNL is as follows:
Multiplier × Number of Contracts × (Delivery Price - Entry Price) = $1 × (-1) × (100 - 200) = $100
Quanto derivatives are a type of exotic derivative. The Quanto architecture is mandatory because I assume the vast majority of market participants wish to deposit USD as margin and withdraw profits in USD, while the underlying spot market's pricing currency is $FLOP. Therefore, we must leverage the Quanto structure to achieve the payoff matrix traders dream of. To price Quanto derivatives with absolute precision, market makers must perform dynamic risk premium adjustments based on the covariance between the FRCR index and the $FLOP/USD exchange rate.
If the market prefers to retain $FLOP directly as the margin currency, the contract design simplifies to:
Contract Multiplier = 1 GigaFLOP
Contract Value = Multiplier × FRCR = 1 × 100 $FLOP = 100 $FLOP
The success or failure of any financial derivatives contract hinges on whether market makers can very conveniently replicate long and short futures positions using the underlying base assets.
Long Futures Position Replication:
To replicate a long exposure, market makers need to acquire $FLOP and purchase a series of sequentially maturing FCUs. Below are the specific steps for a market maker to synthetically construct a 60-day FRCR long futures position:
Spend $FLOP to purchase a 30-day FCU; upon expiration at 30 days, roll the position over into another 30-day FCU.
Assume on Nov 1, the 30-day FCU price is 100 $FLOP; on Nov 30, you sell and settle your original 30-day FCU for 100 $FLOP, while simultaneously purchasing a brand-new 30-day FCU expiring on Dec 30 for 200 $FLOP.
At this point, you have successfully synthetically constructed a 60-day long futures contract, ultimately locking in an entry price of 200 $FLOP.
Short Futures Position Replication:
To replicate a short exposure, market makers need to acquire $FLOP and offer a series of sequentially maturing FCU compute supply contracts. This requires tying up more $FLOP capital, as the staked collateral far exceeds the collected compute fees. Market makers absolutely do not need to physically procure chips, as they will proactively roll over and close positions before expiry/delivery. Below is the specific workflow for synthetically constructing a 60-day FRCR short futures position:
Spend $FLOP as collateral and list a bid to provide a 30-day FCU; prior to the 30-day expiration, roll this position over into another 30-day FCU supply contract.
Assume on Nov 1, the 30-day FCU price is 100 $FLOP; on Nov 30, you buy back your original 30-day FCU for 100 $FLOP to close out the position, and list a new 30-day FCU expiring on Dec 30 for provision at 200 $FLOP. As long as you complete the rollover before the FCU expires, or designate backup miners to fulfill within the 24-hour delivery window, you will never face a default, and your collateral remains intact.
In this way, you perfectly replicate a 60-day short futures contract, with the entry price locked at 200 $FLOP.
FRCR Perpetual Swap
Once you thoroughly understand the operational logic of futures contracts, expanding them into perpetual swaps becomes a natural progression. Remember, a perpetual swap is essentially a continuously rolled-over chain of infinite futures positions.
Perpetual Swap Specifications:
Contract Name: FRCR Quanto Perpetual Swap; Underlying Premium Index: 8-hour Time-Weighted Average Price (TWAP) of [Perpetual Price - FRCR Index].
Perpetual Swap prices will utilize hourly-volume-weighted average prices to strictly match the FRCR calculation methodology; 8 sample points will be collected over an 8-hour cycle for arithmetic averaging.
Contract Multiplier: $1 per point
Example: Assume FRCR Index is at 100 points
Contract Value: Multiplier × FRCR = $1 × 100 = $100
Currency of Margin: USD
The funding rate mechanism operates identically to traditional perpetual swaps. If the premium index is positive, the funding rate is positive, and longs pay shorts; if the premium index is negative, the funding rate is negative, and shorts pay longs. Given this is a Quanto derivative, all funding rate settlements are paid in USD.
To synthetically construct a long perpetual swap, traders simply go long on FCUs and continuously roll over. For instance, if a trader desires a 120-day synthetic long perpetual exposure, they merely purchase a 30-day FCU and roll it over continuously 4 times. Conversely, to synthesize a short perpetual swap, traders simply list bids to supply FCUs and continuously roll over. For example, to establish a 120-day synthetic short exposure, they merely list a supply bid for a 30-day FCU and roll it over continuously 4 times.
AI Model Adaptation Choices
A subtle challenge facing the compute derivatives market lies in the fact that the baseline compute price anchored by financial derivatives is Model-Agnostic; yet in reality, each distinct AI model corresponds to its own independent FCU trading market. Therefore, to truly and accurately synthesize futures and perpetual contracts, traders must purchase or supply basket combinations of FCUs for each independent model according to weight allocations that align with the FRCR composite index.
Alongside publishing the composite index, FRCR will simultaneously release granular data on each model's asset pool in real-time, rendering basket asset weights completely transparent. The composite index will drift as model invocation structures change; for hedgers, the core risk shifts to managing the risk of model iteration and retirement, rather than the physical depreciation of underlying hardware.
Zero-Friction Trading
To cultivate and expand the derivatives market built atop spot and FCU compute prices derived from Flop Network, FCU trades are exempt from any additional platform transaction fees. Both parties involved in a trade only need to pay the network's base $FLOP fee to validators for transaction packaging. This fee bears no difference whatsoever from standard $FLOP transfer fees between wallets.
Core Advantages of the Flop Spot and Forward Markets
Anyone can participate without barriers in compute procurement or compute supply on Flop Network.
All settlement transactions execute within a completely open and transparent marketplace.
All data invocations generated by Flop Network are entirely free.
It is that pure and clear. If you are committed to building any form of compute derivative, all exploration must begin with profound insights into on-chain activity and price discovery within the Flop Network spot and forward markets.
Salute to the pioneers!
[1] RWA: Real World Assets (such as crude oil or gold).