Optimizing Exchange Latency At HFTs I built low-latency trading systems for traditional and cloud-native exchanges including Nasdaq, NYSE Arca, CME, ICE, SGX, JPX, Coinbase/GDAX, and others. A summary/discussion of the current split in low-latency network infrastructure Colocation: Traditional exchanges are hosted in physical datacenters. Colocation involves renting rack space with cross-connects to the exchange matching engine. While it is expensive and time-consuming to set up, physical colocation guarantees an equalized path to the exchange. By contrast, it is fast and inexpensive to set up cloud servers but difficult to guarantee the lowest latency. To be colocated with an exchange running in AWS, firms first need to understand not only what region the exchange is deployed in (e.g. us-east-1) but also the availability zone (e.g. use1-az4). Even within a single availability zone, there is variance in how close a randomly placed EC2 instance lands to the matching engine. Non-determinism: Networking setups for traditional exchanges offer deterministic access to matching engines by running the same cable lengths to all client boxes. Public clouds have a significant degree of non-determinism due to routing within a cloud region. Two different EC2 instances in us-east-1/use1-az4 could have different latencies to the same matching engine in the same zone. Firms mitigate this issue by spinning up multiple instances and choosing the one with lowest ping times, as well as using cluster placement groups and AWS technology such as ENA Express. Unicast instead of multicast: Traditional exchanges disseminate market data over UDP multicast, which puts a single copy of the book on the wire and delivers it to every participant simultaneously. Public clouds don't offer multicast in any form usable for market data: VPCs have no native support, and Transit Gateway multicast is built for enterprise applications rather than microsecond-sensitive feeds. Cloud-native exchanges therefore publish over TCP unicast, which involves a separate copy for every subscriber and a fan-out order that no longer treats them equally. HFTs recover what they can by bypassing the kernel for their networking stack. They use a poll-mode driver such as DPDK over ENA that lets the NIC DMA frames directly into userspace ring buffers, with a userspace TCP stack handling the protocol above it. Inter-region latency: Expensive microwave, millimeter-wave, and subsea fiber links carry the lowest-latency paths between physical datacenters, and none of them terminate inside a cloud region. To move data between cloud regions and traditional venues, firms combine the long-haul routes of established market data vendors with specialized cloud on- and off-ramps, paying a penalty at each transition. Since price discovery still originates largely from traditionally colocated exchanges, trading on public cloud venues adds to a firm's market data infrastructure footprint rather than replacing it. Partnerships between CME and Google and between Nasdaq and Amazon aim to deliver infrastructure that combines the flexibility and global reach of public clouds with the deterministic latency of purpose-built exchange datacenters. If that convergence arrives, most of the practices above stop paying. The advantage in trading on cloud-native exchanges may shift back toward the firms with decades of colocated experience rather than those who have built expertise in cloud-latency engineering.

Architect’s traditional perpetuals exchange AX, one of the fastest-growing regulated derivatives exchanges in history, is opening trading to individuals and omnibus brokers. With $3B notional in <6 months in institutional trading, we’re excited to welcome a new investor class. We launched AX in February with the conviction that building the largest global regulated exchange for traditional asset perpetuals starts with addressing the needs of institutions and professional market participants: robust risk controls, institutional-grade infrastructure, responsible leverage, wash-trading and market manipulation prevention, no internal market maker, and a clear tier-1 regulatory framework. Our GTM strategy was a risk we took based on ambition, industry experience, intuition, and discipline. Our team is now ready to win the opportunities that risk has yielded, in expanding our markets internationally and building what’s needed here in US derivatives markets. AX offers individual traders the same benefits that have enabled our institutional customers to grow the exchange this large this quickly: a market-defining UX, infrastructure built by elite industry experts, superior book liquidity, low fees and open incentive programs, high-throughput and full-featured APIs, competitive margin, 24/7 one-on-one support, no auto-liquidation or ADL. Join us for the next stage of AX and Architect.

We’re hearing from GPU capacity sellers that customers have been rotating from H100s to B200s, but this change isn’t fully reflected in 12-month forward rental prices. Halfway through 2026, the @ComputeDesk H100 index is up ~40% while the newer B200 return sits at ~27%. For most of the year, surging demand for training and interference hit the H100 fleet first while B200 capacity was unfilled. The H100-B200 spread started to close only over the last few weeks in response to a marketwide demand spike.

US Compute Futures As Hedges Compute futures face the argument that if a future doesn’t track the exact costs hedgers incur, it will carry too much basis risk to form a viable market. This fundamentally misrepresents how hedges actually function in many major US markets. A hedge does not need to be perfect in order to be useful, let alone transformative, to its underlying commodity market. To be viable as a multi-billion dollar market, a hedging instrument needs to remove enough risk at scale to be worth its initial cost. If the correlation between a hedging instrument and the portfolio is ρ and the optimal hedge ratio is used, the fraction of variance eliminated by the hedge is ρ². A correlation of only 0.7 cuts variance in half. Cross-hedges built on loose relationships predominate in the real US economy. Airlines hedge jet fuel with crude. Bond desks hedge rate risk in credit with treasuries. Long-short equity portfolios hedge market beta with S&P 500 futures. At my former firms Jane Street and Citadel Securities, every trading desk was required to hedge portfolio factors with related instruments intraday and overnight to isolate alpha. Commodity markets are typically heterogeneous. Compute is not a unique underlying in this respect. An “H100 hour” could represent many different goods: SXM or PCIe, spot or reserved, hyperscaler or neocloud, US or Asia. Weigh this “problem” against the “solution” that compute buyers and sellers have today: nothing. Asset-backed loans on GPUs carry 40-50% haircuts because lenders can’t transfer the associated risks. The institutions financing the >$1T datacenter-linked debt sector regularly transact in other heterogeneous commodity markets. The success of US compute futures/options markets primarily depends on CFTC-regulated exchanges’ agility and competency at collaborating with index providers native to chip configurations, neocloud procurement, and timeseries interpolation on a continuous basis. Designing a futures contract that minimizes basis risk and maximizes liquidity formation is an antecedent requirement. Fulfilling the US government mandate to “accelerate the maturation of a healthy financial market for compute” is the main goal.