The coming futures market in AI compute

The Chicago Mercantile Exchange is set to introduce futures contracts for AI compute, framing it as the new currency for the AI economy. Pete Keavey, the CME executive, stated that these contracts will standardize and trade compute similarly to how oil derivatives evolved. BlackRock CEO Larry Fink anticipates a new asset class will trade these futures.

Boston Consulting Group projects the AI compute market to grow from $360bn in 2025 to $2.3tn by 2030. However, research indicates that two-thirds to three-quarters of new futures contracts fail. For compute derivatives to succeed, prices must be volatile enough for hedging; long lead times and chip bottlenecks, like those affecting Nvidia’s H100 chips, create this volatility.

CME Group, with Silicon Data, has chosen Nvidia’s H100 and B200 chips as references, with rental rates of $2.77 and $5.86 per hour respectively. Contracts will extend 36 months, but standardization remains challenging due to varying performance and lack of consensus on measurement, as seen with competing index providers like Ornn.

The Chicago Mercantile Exchange is set to introduce futures contracts for AI compute, framing it as the new currency for the AI economy. Pete Keavey, the CME executive, stated that these contracts will standardize and trade compute similarly to how oil derivatives evolved. BlackRock CEO Larry Fink anticipates a new asset class will trade these futures.

Boston Consulting Group projects the AI compute market to grow from $360bn in 2025 to $2.3tn by 2030. However, research indicates that two-thirds to three-quarters of new futures contracts fail. For compute derivatives to succeed, prices must be volatile enough for hedging; long lead times and chip bottlenecks, like those affecting Nvidia’s H100 chips, create this volatility.

CME Group, with Silicon Data, has chosen Nvidia’s H100 and B200 chips as references, with rental rates of $2.77 and $5.86 per hour respectively. Contracts will extend 36 months, but standardization remains challenging due to varying performance and lack of consensus on measurement, as seen with competing index providers like Ornn.

The Chicago Mercantile Exchange is set to introduce futures contracts for AI compute, framing it as the new currency for the AI economy. Pete Keavey, the CME executive, stated that these contracts will standardize and trade compute similarly to how oil derivatives evolved. BlackRock CEO Larry Fink anticipates a new asset class will trade these futures.

Boston Consulting Group projects the AI compute market to grow from $360bn in 2025 to $2.3tn by 2030. However, research indicates that two-thirds to three-quarters of new futures contracts fail. For compute derivatives to succeed, prices must be volatile enough for hedging; long lead times and chip bottlenecks, like those affecting Nvidia’s H100 chips, create this volatility.

CME Group, with Silicon Data, has chosen Nvidia’s H100 and B200 chips as references, with rental rates of $2.77 and $5.86 per hour respectively. Contracts will extend 36 months, but standardization remains challenging due to varying performance and lack of consensus on measurement, as seen with competing index providers like Ornn.

The Chicago Mercantile Exchange is set to introduce futures contracts for AI compute, framing it as the new currency for the AI economy. Pete Keavey, the CME executive, stated that these contracts will standardize and trade compute similarly to how oil derivatives evolved. BlackRock CEO Larry Fink anticipates a new asset class will trade these futures.

Boston Consulting Group projects the AI compute market to grow from $360bn in 2025 to $2.3tn by 2030. However, research indicates that two-thirds to three-quarters of new futures contracts fail. For compute derivatives to succeed, prices must be volatile enough for hedging; long lead times and chip bottlenecks, like those affecting Nvidia’s H100 chips, create this volatility.

CME Group, with Silicon Data, has chosen Nvidia’s H100 and B200 chips as references, with rental rates of $2.77 and $5.86 per hour respectively. Contracts will extend 36 months, but standardization remains challenging due to varying performance and lack of consensus on measurement, as seen with competing index providers like Ornn.

The Chicago Mercantile Exchange is set to introduce futures contracts for AI compute, framing it as the new currency for the AI economy. Pete Keavey, the CME executive, stated that these contracts will standardize and trade compute similarly to how oil derivatives evolved. BlackRock CEO Larry Fink anticipates a new asset class will trade these futures.

Boston Consulting Group projects the AI compute market to grow from $360bn in 2025 to $2.3tn by 2030. However, research indicates that two-thirds to three-quarters of new futures contracts fail. For compute derivatives to succeed, prices must be volatile enough for hedging; long lead times and chip bottlenecks, like those affecting Nvidia’s H100 chips, create this volatility.

CME Group, with Silicon Data, has chosen Nvidia’s H100 and B200 chips as references, with rental rates of $2.77 and $5.86 per hour respectively. Contracts will extend 36 months, but standardization remains challenging due to varying performance and lack of consensus on measurement, as seen with competing index providers like Ornn.

The Chicago Mercantile Exchange is set to introduce futures contracts for AI compute, framing it as the new currency for the AI economy. Pete Keavey, the CME executive, stated that these contracts will standardize and trade compute similarly to how oil derivatives evolved. BlackRock CEO Larry Fink anticipates a new asset class will trade these futures.

Boston Consulting Group projects the AI compute market to grow from $360bn in 2025 to $2.3tn by 2030. However, research indicates that two-thirds to three-quarters of new futures contracts fail. For compute derivatives to succeed, prices must be volatile enough for hedging; long lead times and chip bottlenecks, like those affecting Nvidia’s H100 chips, create this volatility.

CME Group, with Silicon Data, has chosen Nvidia’s H100 and B200 chips as references, with rental rates of $2.77 and $5.86 per hour respectively. Contracts will extend 36 months, but standardization remains challenging due to varying performance and lack of consensus on measurement, as seen with competing index providers like Ornn.

The Chicago Mercantile Exchange is set to introduce futures contracts for AI compute, framing it as the new currency for the AI economy. Pete Keavey, the CME executive, stated that these contracts will standardize and trade compute similarly to how oil derivatives evolved. BlackRock CEO Larry Fink anticipates a new asset class will trade these futures.

Boston Consulting Group projects the AI compute market to grow from $360bn in 2025 to $2.3tn by 2030. However, research indicates that two-thirds to three-quarters of new futures contracts fail. For compute derivatives to succeed, prices must be volatile enough for hedging; long lead times and chip bottlenecks, like those affecting Nvidia’s H100 chips, create this volatility.

CME Group, with Silicon Data, has chosen Nvidia’s H100 and B200 chips as references, with rental rates of $2.77 and $5.86 per hour respectively. Contracts will extend 36 months, but standardization remains challenging due to varying performance and lack of consensus on measurement, as seen with competing index providers like Ornn.

The Chicago Mercantile Exchange is set to introduce futures contracts for AI compute, framing it as the new currency for the AI economy. Pete Keavey, the CME executive, stated that these contracts will standardize and trade compute similarly to how oil derivatives evolved. BlackRock CEO Larry Fink anticipates a new asset class will trade these futures.

Boston Consulting Group projects the AI compute market to grow from $360bn in 2025 to $2.3tn by 2030. However, research indicates that two-thirds to three-quarters of new futures contracts fail. For compute derivatives to succeed, prices must be volatile enough for hedging; long lead times and chip bottlenecks, like those affecting Nvidia’s H100 chips, create this volatility.

CME Group, with Silicon Data, has chosen Nvidia’s H100 and B200 chips as references, with rental rates of $2.77 and $5.86 per hour respectively. Contracts will extend 36 months, but standardization remains challenging due to varying performance and lack of consensus on measurement, as seen with competing index providers like Ornn.

The Chicago Mercantile Exchange is set to introduce futures contracts for AI compute, framing it as the new currency for the AI economy. Pete Keavey, the CME executive, stated that these contracts will standardize and trade compute similarly to how oil derivatives evolved. BlackRock CEO Larry Fink anticipates a new asset class will trade these futures.

Boston Consulting Group projects the AI compute market to grow from $360bn in 2025 to $2.3tn by 2030. However, research indicates that two-thirds to three-quarters of new futures contracts fail. For compute derivatives to succeed, prices must be volatile enough for hedging; long lead times and chip bottlenecks, like those affecting Nvidia’s H100 chips, create this volatility.

CME Group, with Silicon Data, has chosen Nvidia’s H100 and B200 chips as references, with rental rates of $2.77 and $5.86 per hour respectively. Contracts will extend 36 months, but standardization remains challenging due to varying performance and lack of consensus on measurement, as seen with competing index providers like Ornn.

The Chicago Mercantile Exchange is set to introduce futures contracts for AI compute, framing it as the new currency for the AI economy. Pete Keavey, the CME executive, stated that these contracts will standardize and trade compute similarly to how oil derivatives evolved. BlackRock CEO Larry Fink anticipates a new asset class will trade these futures.

Boston Consulting Group projects the AI compute market to grow from $360bn in 2025 to $2.3tn by 2030. However, research indicates that two-thirds to three-quarters of new futures contracts fail. For compute derivatives to succeed, prices must be volatile enough for hedging; long lead times and chip bottlenecks, like those affecting Nvidia’s H100 chips, create this volatility.

CME Group, with Silicon Data, has chosen Nvidia’s H100 and B200 chips as references, with rental rates of $2.77 and $5.86 per hour respectively. Contracts will extend 36 months, but standardization remains challenging due to varying performance and lack of consensus on measurement, as seen with competing index providers like Ornn.

The Chicago Mercantile Exchange is set to introduce futures contracts for AI compute, framing it as the new currency for the AI economy. Pete Keavey, the CME executive, stated that these contracts will standardize and trade compute similarly to how oil derivatives evolved. BlackRock CEO Larry Fink anticipates a new asset class will trade these futures.

Boston Consulting Group projects the AI compute market to grow from $360bn in 2025 to $2.3tn by 2030. However, research indicates that two-thirds to three-quarters of new futures contracts fail. For compute derivatives to succeed, prices must be volatile enough for hedging; long lead times and chip bottlenecks, like those affecting Nvidia’s H100 chips, create this volatility.

CME Group, with Silicon Data, has chosen Nvidia’s H100 and B200 chips as references, with rental rates of $2.77 and $5.86 per hour respectively. Contracts will extend 36 months, but standardization remains challenging due to varying performance and lack of consensus on measurement, as seen with competing index providers like Ornn.

芝加哥商品交易所将推出AI算力期货合约，将其定义为AI经济新货币。Pete Keavey，CME执行官表示，这些合约将像石油衍生品一样标准化和交易算力。BlackRock首席执行官Larry Fink预计将交易这些衍生品的新资产类别将出现。Boston Consulting Group预测，AI算力市场将从2025年的3600亿美元增长至2030年的2.3万亿美元。然而，研究表明，多达三分之二到四分之三的新期货合约会失败。为了使算力衍生品成功，价格必须足够波动以便对冲；长交期和芯片短缺，如影响Nvidia H100芯片的问题，会产生这种波动。CME Group与Silicon Data合作，将Nvidia的H100和B200芯片作为参考，每小时租金分别为2.77美元和5.86美元。合约将延期36个月，但由于性能各异且缺乏共识的计量标准，标准化仍面临挑战，正如Ornn等竞争指数提供商所见。