Brett Harrison (@BrettHarrison) header
Brett Harrison (@BrettHarrison) profile picture

Founder & CEO @Architect_Fi, ex-Citadel Securities/Jane Street, dev, dad

Joined May 2021
Photo by Brett Harrison (@BrettHarrison)Photo by Brett Harrison (@BrettHarrison)Photo by Brett Harrison (@BrettHarrison)Photo by Brett Harrison (@BrettHarrison)Photo by Brett Harrison (@BrettHarrison)Photo by Brett Harrison (@BrettHarrison)Photo by Brett Harrison (@BrettHarrison)Photo by Brett Harrison (@BrettHarrison)Photo by Brett Harrison (@BrettHarrison)Photo by Brett Harrison (@BrettHarrison)
Introducing the American Innovation Exchange, the first U.S. derivatives exchange designed for trading the AI economy. Trade futures and options on compute, metals, energy, and other critical instruments in the AI supply chain. Coming soon from Architect.
42
53
10
444
111,323
Brett Harrison retweeted
We compared Liquid Inference with 8 routers and gateways. Each page has a head-to-head table, a fee calculator and dated sources. What the exchange does differently: providers compete for each request, and every request can have a hard cap. liquidinference.ai/compare
1
4
1
12
1,250
Brett Harrison retweeted
The enterprise playbook won't do much to control costs. Only the expensive products will respond to RFPs. It's not flexible enough for the current market. Liquid Inference is the jack-of-all trades for AI procurement. The top providers are constantly juggling their unused compute. They'll be serving one month and then fully booked with dedicated contracts the next. It's like herding cats. What they should do is run tests with a tight budget to see where you can substitute to cheaper models. Then hit up our entire provider network for a quote based on your specs. We'll herd the cats and fill in all the blanks.
Reserve capacity on Liquid Inference. Pick a model (text, video, image or speech) and set your terms: a monthly pool, rate limits, uptime, region and a term of up to 12 months. You get a quote, and you can negotiate before you accept. Fixed monthly fee. Zero data retention on every reservation. inference.ai.exchange/docs/r…
3
11
1,005
We've steadily improved our Liquid Inference routing algorithm to optimize for cache hit-rate. We've achieved >90% average across models, often realizing higher rates with DeepSeek and GLM providers.
14
10
64
4,210
Brett Harrison retweeted
Liquid Inference checks every tool call a provider returns. Bad call on a non-streamed request? We drop it and send the job to the next offer. Provider returns nothing? You pay nothing. Last 7 days: 99.64% of tool-call answers had no bad call. liquidinference.ai/docs/erro…
6
2
22
1,516
Brett Harrison retweeted
Reserve capacity on Liquid Inference. Pick a model (text, video, image or speech) and set your terms: a monthly pool, rate limits, uptime, region and a term of up to 12 months. You get a quote, and you can negotiate before you accept. Fixed monthly fee. Zero data retention on every reservation. inference.ai.exchange/docs/r…
1
5
2
19
2,243
Mean reversion in the GPU rental market. Over the past month, 1-year US GPU rental rates for A100, H100, and H200 have reverted back to mid-October levels, while B300 has given back most of its late-October gains. What we think is driving the divergence: - Neocloud credit pressure. With neocloud equities down sharply and borrowing costs rising, some operators are moving B300 capacity at a discount to raise cash. - Inference is driving demand. Neocloud demand continues to shift from training to inference (Bloomberg, below). Better small-to-mid-size open-weight models make older Nvidia accelerators more economical for many common inference workloads. - B300 supply is catching up. B300s are the most commonly requested chip in the OTC market, and availability has historically been scarce. But we now see a number of B300/GB300 clusters coming online between end of Q4 2026 and mid Q1 2027. We're brokering spot and options trades across the accelerator spectrum — reach out if interested.
21
26
9
150
15,458
Brett Harrison retweeted
In Chicago for SC26? Join us at Fioretta on Tuesday, Nov 17 with CMT Digital and Luxor. Bring your questions on trading compute and inference. RSVP below.
1
3
13
1,770
Brett Harrison retweeted
Bring your own key. Connect your own Anthropic, OpenAI, Bedrock, Vertex AI or Azure account (16 clouds) and send requests through the same Liquid Inference API. Your cloud bills you at your own rates. If your account can't serve a request, the market takes it before the first token. You can also set it to fail instead. inference.ai.exchange/docs/b…
1
5
1
17
1,433
Brett Harrison retweeted
Gate Ventures Podcast | Episode 19 Compute as the Next Commodity: From GPU Futures to Inference Markets Join Gate Ventures as we sit down with Andrew Lee, Chief Technology Officer at @Architect_Fi, Welcome to our live show! xitter.gq/i/broadcasts/2LPXbgMML…
82
88
4
132
5,894
Brett Harrison retweeted
Our CTO @chairmanlee_ joined the Gate Ventures pod to discuss how Architect is shaping the future of compute capacity and inference markets. Enjoy and follow Andrew!
Gate Ventures Podcast | Episode 19 Compute as the Next Commodity: From GPU Futures to Inference Markets Join Gate Ventures as we sit down with Andrew Lee, Chief Technology Officer at @Architect_Fi, Welcome to our live show! xitter.gq/i/broadcasts/2LPXbgMML…
1
5
1
19
1,385
Brett Harrison retweeted
AI infra credit crunch, CFTC comment period on compute derivs, GPU rental prices up, inference tokens down. Now a blue wave that threatens to slow the build out. I can feel the space tighten between the rock and a hard place. We need the compute spark spread.
3
27
1,249
Brett Harrison retweeted
Kimi K3.1 Pro is on the Liquid Inference exchange. Providers post offers, and each request goes to the lowest one that meets your terms. As more providers list it, they compete for your requests. $2.00 in · $0.20 cached · $10 out, per 1M tokens. liquidinference.ai
6
7
2
34
48,221
It’s compute day at FIA Chicago. Index providers are up first at 1:30pm, exchanges offering compute futures to follow at 2:20pm. Compute Desk, Silicon Data, NativX, Ornn, Architect, CME, ICE, and Nodal all in one place. Expect new alpha to be dropped.
4
7
114
7,261
Brett Harrison retweeted
Hit us up if you’re gonna be at FIA Expo this year.
Superintelligence markets will begin in Chicago due to DRW, Jump, CME, Architect, and hundreds of CFTC-regulated industry partners working to financialize compute and inference in the US. If you’re at FIA come visit Architect’s office, trade B300s, and play our CTO in 301.
2
4
13
1,112
Brett Harrison retweeted
Gate Ventures Podcast | Episode 19 Compute as the Next Commodity: From GPU Futures to Inference Markets Could GPU capacity become the next tradable commodity as AI compute demand continues to surge? Join Gate Ventures as we sit down with Andrew Lee, Chief Technology Officer at @Architect_Fi, to explore: 🔹 How GPU capacity could evolve into a standardized, tradable commodity 🔹 The rise of GPU futures and the challenges of compute delivery markets 🔹 How compute hedging could reshape financing for neoclouds and AI infrastructure 🔹 The future of inference markets and who stands to benefit from compute price volatility 📅 November 5 | 15:00 UTC | 23:00 GMT+8 📍 Video livestream on Gate Ventures' X — also live on X Spaces.
73
78
103
4,594
A story about how Jane Street unwittingly reinvented datacenters in 2012, involving office server rooms, natural disasters, market maker obligations, recovery sites, colocation facilities, and old limestone telco buildings in lower Manhattan. When I joined Jane Street from college in 2010, all of the firm’s proprietary trading systems ran out of a 8’X12’ server room on the 33rd floor of 1 New York Plaza, on the southern tip of Manhattan. Jane Street’s ETF market-making, options vol pricing, intercountry ADR arbitrage, and pan-equity stat-arb operated from this small deployment. The room itself was an engineering feat: the Dell server arrays were too heavy for the 1960s building’s floors to support, so the racks were affixed to the ceiling with concrete. By 2012 Jane Street was not yet a member of most major US exchanges, even NYSE, and had virtually no presence in colocation facilities. Nevertheless, it was a ~$2B company experiencing rapid growth, a designated liquidity provider in most listed ETFs, a significant percentage of US and international equity volume, and a critical operator in capital markets. In October 2012, Hurricane Sandy devastated lower Manhattan. 1 New York Plaza, Jane Street’s home, flooded to street level and the basement power generators were fully submerged. The resulting gas leaks made the office building hazardous to enter, and power and water were weeks away from being restored. Unprecedentedly, all major US exchanges closed for two business days after the storm but would be back up by day three. Jane Street was in danger of failing its lead-market-maker obligations if it didn’t relocate its business in time. Jane Street had recently set up a partial office in a central New Jersey office park to comply with FINRA/SEC regulatory obligations for business continuity planning. A few engineers and I showed up to the empty site during the hurricane to begin mass-redeploying the firm’s software and infrastructure for the next critical 48 hours. A truckload of Dell PowerEdge R610s arrived the next morning, some salvaged from 1 New York Plaza’s server room, carried down 33 floors in pitch black by Jane Street’s IT team. For the next few months, everyone at the company from Rob Granieri to new-hires rode rented regional travel buses from Manhattan to NJ to keep Jane Street alive and growing. When we all returned to Jane Street’s repaired 1 New York Plaza office in January 2013, the technology team realized something that’s obvious now: hardware doesn’t need to be colocated with people. The NJ disaster site had accidentally become Jane Street’s first firm-wide datacenter. The site afforded significantly more space, cheaper electricity, ground-floor weight support, and abundant cooling. This decision, to leave Jane Street’s servers in post-hurricane locations, would have profound effects on the firm’s infrastructure footprint and growth. In 2013 it became a top priority for the technology and infrastructure teams to move all latency-sensitive trading into colocation facilities in order to cross-connect directly to exchange matching engine switches. And the search began for a new primary datacenter to house the firm’s research cluster and non-latency-sensitive applications. In 2015 Jane Street found its new electronic home: lower Manhattan, limestone and 32-story windows, originally owned by the New York Telephone Company. It had retina-scan security and neon-colored liquid cooling and everything else we know today. They joined other major financial firms and software companies who had operated there for years. In modeling disaster scenarios, experts tend to underrate the technological redundancies that already exist, or can be willed into rapid deployment, to ensure the US financial system remains secure.
29
57
6
812
60,829
Brett Harrison retweeted
spare some tokens on Liquid Inference so decided to craft this video for any folks interested to understand these topics without background knowledge. here's the prompt: "Using Liquid MCP, generate a video for X social post to explain the below article for people with no background in ML" around 0.5$ in total
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training. We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵 Blog: rikiyat.github.io/blog/weigh…
Readers added context they thought people might want to know
LLM was given weights explicitly as text in prompt and used code/forward passes (per authors' logs) to ablate Qwen looping direction, a known thinking-mode issue per model card. Primarily fixed repetition in that mode, not general 'natively' improving weights/math. rikiyat.github.io/blog/weight-sp… github.com/RikiyaT/lm_und… huggingface.co/Qwen/Qwen3.5-0… huggingface.co/RikiyaT/Qwen3.…
2
2
12
1,125
Superintelligence markets will begin in Chicago due to DRW, Jump, CME, Architect, and hundreds of CFTC-regulated industry partners working to financialize compute and inference in the US. If you’re at FIA come visit Architect’s office, trade B300s, and play our CTO in 301.
5
9
3
107
7,687