# CEO, Compute Exchange — Carmen Li — session 2026-10-14T16:20:00.000Z → 2026-10-14T16:38:00.000Z

_53 transcript lines · 54 slides · source: full recording_

## Transcript

**[00:00:11 · 0]** Alright. Happy Thursday. I like how full this room is. It's Wednesday. It's Wednesday?

**[00:00:18 · 0]** It's Wednesday. Oh, sorry. I'm clearly ahead of everybody here. Happy Wednesday. My name is Carmen Lee.

**[00:00:25 · 0]** I'm the CEO for Compute Change as well as Silicon Data. I love the, you know, the previous session. We talked... We talked about a lot about the whole ecosystem, the financiers, and the spot market traders, and the hedgers. So what we're sitting in is really cool.

**[00:00:43 · 0]** I have a company called Compute Change. The world's actually the only neutral marketplaces for GPU procurement. We started in 2024 when there were, I don't know, five new clouds, three hyperscalers. Essentially, we sort of will foresee, interestingly enough, the proliferation of the GPU ecosystem. So you think about stock market today.

**[00:01:08 · 0]** Right? When you try to buy and sell stocks, you can call brokers. Obviously, you can call a few banks to say, hey. I wanna buy and sell x amount of Tesla or actually SpaceX. Right?

**[00:01:19 · 0]** And you can do that, or you can transact through your app at Nicey or Nasdaq. Doesn't mean you get a best price at Nicey or Nasdaq just because it's more efficient, and you know the liquidity will be there. You know the trade will be settled. It'll be cleared. You can get what you're looking for and the prices you're looking for in a very efficient manner.

**[00:01:40 · 0]** Right? For capital markets and resource allocation, it's about efficiency. So in compute world, obviously, now we have a huge proliferation, which is great. We have over 300 new cloud private provider globally. We have a lot of token factories.

**[00:01:56 · 0]** We have a lot of API routers, which is all good for end customers. But at the same time, introduce a very interesting variable is how do we make transaction efficient? Right? Do you wanna call 300 salespeople? I mean, it's great to make friends.

**[00:02:13 · 0]** Right? You want to talk to them regular basis. They know a lot of things. But same time, when you need compute in the next forty eight hours or reserve contracts, it's the easiest way is go to an exchange and then go to the verified compute resources, just like go to Nasdaq and IC to buy stocks. Right?

**[00:02:30 · 0]** So it's one of the things computing does extremely well is we have this neutral platform. And by design, we do not own GPUs. Once we start owning GPUs, that would translate to... Just like Nasdaq owning stocks, translate to I have a conflict of interest with you and with the providers. I could cherry pick clients.

**[00:02:52 · 0]** Right? I could hold relationships because the compute change does not own anything by design. We work really well with our providers globally. And one very important thing is we treat every single transaction individually, meaning we want to find you as end customer the most suited compute resources. And we focus on reserve compute, forward contracts, as well as token packages.

**[00:03:21 · 0]** So give you some high level. So today, we have over 100 verified providers, and over 80,000 GPU in supply. We cover all regions, obviously ex China. Our commitment is twenty four hours turnaround. The difference between us and other so called brokers is we do verify GPUs.

**[00:03:50 · 0]** If it's existing reserve capacity, meaning things are actually in a functional stage, we'll verify the UIDs, the inventories, the performances, we run Nvidia DGX, InferenceX, MLPerf, make sure you get resources you're promised to get. So we're the middle middle sort of trusted platform. So if it's full capacity, let's say you need GPU starting 01/01/2027, We actually look through the purchase orders, all the permits they get, make sure, you know, we help you do the due diligence level. Now on the flip side, right, if you're a new cloud, your concern is there's a zillion clients looking for PinPew. How do you know who has the ability to pay at a scheduled rate?

**[00:04:35 · 0]** That's what Computers comes in. We do KYB, KYC of the clients. Obviously, you wanna make sure your clients is the clients you think they who they are, So we step in and do that kind of verification just like any exchange would do. And everything we do in the four markets is actually the precursor to physically deliver futures. I'll also give you some context why I am...

**[00:04:58 · 0]** What I do. Sorry. I probably should say that. I start my career as energy option market maker, at DRW Trading in Chicago. So I did that for four or five years.

**[00:05:07 · 0]** So as option... Energy market maker, everything's very intuitive. You have double TI, you have brand. Right? Double TI is a physically delivered commodity contracts.

**[00:05:17 · 0]** Brand is cash settled. They also have very distinct use cases. If you're looking for hedging in the financial sense, right, your price volatility, your cost volatility, you're going to brand, You're looking to physically procure resources, use w t I. So we're creating the w t I at compute change. So the ways to buy compute through, compute change right now, you can get reserved contracts.

**[00:05:47 · 0]** You can get forward contracts on both GPU and token. Let me explain. The forward contracts on GPU is very reasonable. Right? You want something starting in the future, nothing available right now.

**[00:05:57 · 0]** So that's something we're doing on regular basis. Token forward, I think, can be a new ish concept. Basically, you can think about as a dedicated deployment, on premise deployment. Right? Similarly, you reserve a bare metal GPU clusters.

**[00:06:13 · 0]** You reserve a cluster, but you have dedicated token deployment on top of that. So you get guaranteed, SLAs on that particular model. You get, quantization, precision. Everything's fixed for you for a period of time. So we try to solve the on demand token API volatility pricing problem here.

**[00:06:35 · 0]** The third is we have a huge market, open hardware market. It's really interesting. The secondary server transaction started last year when people just pinging me and say, hey, Carmen, I have like 20,000 h 100 sitting x data center. Can you take away from me? So I was like, I don't know.

**[00:06:54 · 0]** I can't store in my home, obviously. So we started having the marketplace for secondary server sales. This has been a very vibrant market, for the past, I would say, twelve months. We cover again all servers, even b 200, you see secondary transaction is happening, sometimes trading above MSRP. Right?

**[00:07:14 · 0]** Can make sense. Think about, like, COVID times when secondary car is trading above. Right? It's a brand new car price. So we cover the three major layers of compute.

**[00:07:28 · 0]** So reserve GPU, obviously, I think the previous speaker covered extremely well with the terms. The minimum we do is one node starting at one month, and then we can go all the way to three, four, five years. Right? It all depends on the availability, the providers, but you get a chance to state your preferences. We don't do on demand.

**[00:07:51 · 0]** And another thing really works really well for compute change is we do provide a flexibility. Let's say if you over commit in terms of quantity or duration, you can sell it back to compute change. Right? So that's something we're coming to do, so you don't have to worry too much if you overbought. Alright.

**[00:08:12 · 0]** Segment live. So I will actually do a quick demo later. So we have a lot of listings, marketplace listings. You can browse and chat with the providers and figure out the best delivery mechanisms. You can also RFQL if you don't see anything you like...

**[00:08:26 · 0]** You liked. We're actually seeing over a billion RFQL every month. Sounds big, but we all know it's full compute. It's not that huge, right? But still you see the the the the the activities, which I love is we see a big range of rf kill requests from one node all the way to 20,000 nodes.

**[00:08:48 · 0]** So our clients are, you know, all different shapes and forms, different use cases, which is great. Yeah. So you set timing, submit, and we route it. So in the behind the scene, we have inventory tracking. So with over a 100 providers globally, we know who has what, roughly when, right, the forward capacity.

**[00:09:08 · 0]** And then let's say you you submit a request in that, hey. I need 30, 300 GPUs, b 200, USC starting January 1. We say, you know, we have three providers. We think they have those. We'll route it to the three.

**[00:09:21 · 0]** They will respond to you through the platforms, and it'll give you, a rubric spec. Pricing, specs, SLAs, we benchmark them if it's existing servers, and then it's your choice, right? Price is not the only indicator. We believe give you as much transparency as we can, and then you will make a decision. So I record in one place.

**[00:09:44 · 0]** Again, so rubrics coming back. You will see everything, and we introduce you to each other. We never hold relationships. Right? We'll make sure you know who you're dealing with.

**[00:09:54 · 0]** So by tomorrow's computer, today's price. So it's a physically procured platform. This is the used refurbished, GPUs. So again, we have extremely, exciting ecosystem participants in this eco... In this, secondary GPU cells.

**[00:10:13 · 0]** We cover from four NIOs to b 200 AMD, everything. Inventory is live, and we do connect with you through all different channels, Slack, email, chat, however you wanna do this. We track end to end. We actually have a scoring for everybody. We score you, our customers.

**[00:10:36 · 0]** We score our providers. You can't see it quite yet, but, you know, it will be really fun. We're gonna have end of year celebration for our providers. We can we're gonna give out the best provider in APAC, best provider in Europe, best provider in US, most responsive providers. So we're gonna have a lot of provider awards coming up.

**[00:10:55 · 0]** Stay tuned. Verify network. So everything's, you know, verified, which is, you know, negotiated exchange. So we help in terms of be the middleman. Sometimes things are not very clear.

**[00:11:09 · 0]** We try to make sure people understand what's the best practices we have seen in the ecosystem. I'm gonna quickly show you guys the platform. I think that's more fun. Here. So this is the platform.

**[00:11:35 · 0]** You can see a bunch of available inventories. By the way, anyone can can sign up. It's completely free. There's no downside using us. You don't have to buy.

**[00:11:42 · 0]** You can browse. You You can request for quote if you, you know, interest in something we don't have. You can start with your vendor type. Again, we're agnostic of GPUs. You can select your model.

**[00:11:53 · 0]** You can select your deployment type. You can select number of GPUs. You can select your computer connect preferences, CPU platforms, storages, everything you need, and then some option items with systems, software stacks, regions. Obviously, we encourage you to select multiple regions if you're open to that. You can add head notes, tell us your use cases, anything you want, and then submit an RFQ.

**[00:12:20 · 0]** So this has been extremely robust, and then we, you know, we obviously, we wanna make sure, if you have any feedback, let us know. So I'm gonna pause on the Compute Change side. I think I have some time left, which I'd love to show you some of the numbers we see in terms of term structure. Also, context, I'm CEO for Compute Change. I'm also CEO for another company called Silicon Data.

**[00:12:46 · 0]** So I'm a founder CEO for Silicon Data, which is the Bloomberg for Compute. So we provide indices for CME to settle their futures on, which will be happening in a few weeks. I'm very excited about this. So the ones that's gonna start trading at CME will be the h 100 NeoCloud on demand rates as well as v 200, prices. What I want to show you is implied forward curve.

**[00:13:14 · 0]** So as I said, Confucius will do a lot of forward capacities. So essentially, a lot of longer term contracts is gonna be very critical. You think about even in your cloud or your your consumers. Let me just make it larger. So the yellow line was last year.

**[00:13:32 · 0]** Right? The the x horizontal line is the duration, and the dollar amount is the y axis. So last year was complete downward sloping curve from the yellow line. Right? So meaning the longer term contract you signed, the cheaper, you know, was per GBP hour.

**[00:13:49 · 0]** And this year, starting early this year, the whole curve shift up and became contango. If you see the the red line, right? So it translates to people not getting a discount, enough discount for a longer term contract implies that people assume the price is gonna rise down the line. And now you see the price the whole term keeps shifting up but got flattened, right? So meaning people will now either get penalized or get, you know, awarded by signing long term contract.

**[00:14:19 · 0]** This is reserve contract we observe, and if you look at implied fall rates, this is not this is implied fall rates, right, because we don't have a future trading. So this is calculated based on our term structure, and we use similar methodology at treasury futures curve. So you can see a market consensus view is the price could go up next in the future, next thirty six months. Again, market consensus doesn't mean it's right. If you look at b 200, again, last year was contango, and this year, the whole curve shifted up.

**[00:14:53 · 0]** Every single term is more expensive, and it's got flattened. Right? Implied forward rates, similarly, people think the price is gonna go up. I want to point this out, the term structure out, is this shows you that GPU behaves just like every commodity, have a very clear term structure risks, which data futures are excellent at hedging. So we have seen a lot of block trading happening around the term structure.

**[00:15:20 · 0]** So you think about this, right? If you have servers on your balance sheet, you have financing obligation for the next five, six years, and you have reserved contracts for first three. Then they're exposed essentially year four, five, or six. Right? So you want to hedge that risk out by entering, you know, swaps or block trading.

**[00:15:41 · 0]** So those are things... Or or there are there are trades happening even for the first three years, which you have reserved contracts on. What if the reserve contract defaulted? Right? What's your guarantee from some other players.

**[00:15:56 · 0]** So we saw block trading happening around the indices, trading as a floating rate. This chart, actually is being used differently by NVIDIA team. In NVIDIA's latest September investor day deck, they set us with the only data provider that actually cited in their deck. This is very interesting. Slick and data becomes the appraisal for a few banks when they have GPU servers on their balance sheet.

**[00:16:32 · 0]** This, everything you see here, a forward looking view, right? So we discount the cash flow. So as I said, we have a term structure, so assume A100s, right? A100s have been around for six years. If today you sign a three year long lease on A100s and you have a rate, and we discount the cash back to time zero, you have how much the asset is worth.

**[00:16:55 · 0]** Right? Obviously, there's a lot of things baked in, utilization rate, what is your cost. There's a bunch of assumptions which we're pretty transparent about. We actually calculate the residual value for normalized value obviously across all a 100, h 100, b 200 servers globally, and then we'll compare that with the straight line depreciation, which is everybody using the five year depreciation. What's interesting is the a 100.

**[00:17:22 · 0]** Right? Even though it is downward sloping, you know, it means these things are depreciating, but it's above the accounting depreciation curve straight line. And h 100 obviously is, you you know, it's going going bananas now. The price actually went up. My my analogy will be just like secondary cars during COVID time.

**[00:17:42 · 0]** The price went up, right? And b 200 obviously went above the MSRP by this point. So this is give some... Give you some views about what's going on in the Residue Valley side. I think I'm out of time, but, you know, feel free to grab me, Carmen, cli at

## Slides

### 00:00:00–00:00:20

# AI Engineer
## NEW YORK

### 00:00:20–00:00:40

# Compute Exchange
Product presentation · October 2026

The neutral GPU marketplace for AI infrastructure. Reserved GPUs, token & forward contracts, and hardware across 100+ verified providers.

compute.exchange
Product overview

### 00:00:40–00:01:00

# Compute Exchange
- Product presentation - October 2026

The neutral GPU marketplace for AI infrastructure. Reserved GPUs, token & forward contracts, and hardware across 100+ verified providers.

[Logo for Compute Exchange]

### 00:01:00–00:01:20

# Compute Exchange
- Product presentation • October 2026

The neutral GPU marketplace for AI infrastructure. Reserved GPUs, token & forward contracts, and hardware across 100

### 00:01:20–00:01:40

## Compute Exchange
The neutral GPU marketplace for AI infrastructure. Reserved GPUs, token & forward contracts, and hardware across 100+ verified providers.
compute.exchange

[Logo for Compute Exchange]

### 00:01:40–00:02:00

# Compute Exchange
- Product presentation • October 2026
The neutral GPU marketplace for AI infrastructure. Reserved GPUs, token & forward contracts, and hardware across 100+ verified providers.

[Green logo for Compute Exchange]

### 00:02:00–00:02:20

# Compute Exchange
Product presentation · October 2026
The neutral GPU marketplace for AI infrastructure. Reserved GPUs, token & forward contracts, and hardware across 100+

### 00:02:20–00:02:40

# Compute Exchange
- Product presentation • October 2026

The neutral GPU marketplace for AI infrastructure. Reserved GPUs, token & forward contracts, and hardware across 100+ verified providers.

[Logo with interlocking shapes for Compute Exchange]

### 00:02:40–00:03:00

# Compute Exchange
The neutral GPU marketplace for AI infrastructure. Reserved GPUs, token & forward contracts, and hardware across 100+ verified providers.

[Logo for Compute Exchange, featuring

### 00:03:00–00:03:20

# Compute Exchange

- Product presentation • October 2026

The neutral GPU marketplace for AI infrastructure. Reserved GPUs, token & forward contracts, and hardware across 100+ verified providers.

compute.exchange

[Logo for Compute Exchange]

### 00:03:20–00:03:40

## Compute Exchange
- Product presentation • October 2026

The neutral GPU marketplace for AI infrastructure. Reserved GPUs, token & forward contracts, and hardware across 100+ verified providers.

compute.exchange
Product overview

[Logo for Compute Exchange, featuring a green square icon]

### 00:03:40–00:04:00

### The exchange
# Compute will become a commodity. We are building its marketplace.

Compute Exchange connects AI teams with verified GPU providers. We stay neutral and do not operate compute ourselves. Providers compete on your requirement and you receive one normalized quote.

- **100+** Verified providers
- **80K+** GPUs in supply
- **6** Regions - US / EU / APAC
- **24h** Quote turnaround

### Keynote
Carmen Li / CEO compute.exchange

### 00:04:00–00:04:20

The exchange
# Compute will become a commodity. We are building its marketplace.

Compute Exchange connects AI teams with verified GPU providers. We stay neutral and do not operate compute ourselves. Providers compete on your requirement and you receive one normalized quote.

- **100+** Verified providers
- **80K+** GPUs in supply
- **6** Regions - US / EU / APAC
- **24h** Quote turnaround

### 00:04:20–00:04:40

### The exchange
# Compute will become a commodity. We are building its marketplace.

Compute Exchange connects AI teams with verified GPU providers. We stay neutral and do not operate compute ourselves. Providers compete on your requirement and you receive one normalized quote.

- **100+** Verified providers
- **80K+** GPUs in supply
- **6** Regions - US / EU / APAC
- **24h** Quote turnaround

## Keynote
Carmen Li / CEO
compute. exchange
[Logo for Compute Exchange, showing a stylized 'C' and 'E' forming a square shape]

### 00:04:40–00:05:00

# Compute will become a commodity. We are building its marketplace.

The exchange

Compute Exchange connects AI teams with verified GPU providers. We stay neutral and do not operate compute ourselves. Providers compete on your requirement and you receive one normalized quote.

- **100+** Verified providers
- **80K+** GPUs in supply
- **6** Regions - US / EU / APAC
- **24h** Quote turnaround

Keynote
Carmen Li / CEO compute.exchange

### 00:05:00–00:05:20

### The exchange
# Compute will become a commodity. We are building its marketplace.

Compute Exchange connects AI teams with verified GPU providers. We stay neutral and do not operate compute ourselves. Providers compete on your requirement and you receive one normalized quote.

- **100+** Verified providers
- **80K+** GPUs in supply
- **6** Regions - US / EU / APAC
- **24h** Quote turnaround

## Keynote
Carmen Li / CEO
compute.exchange

### 00:05:20–00:05:40

### The exchange
## Compute will become a commodity. We are building its marketplace.

Compute Exchange connects AI teams with verified GPU providers. We stay neutral and do not operate compute ourselves. Providers compete on your requirement and you receive one normalized quote.

-   **100+** Verified providers
-   **80K+** GPUs in supply
-   **6** Regions - US / EU / APAC
-   **24h** Quote turnaround

### 00:05:40–00:06:00

### Product lines
# Three ways to buy compute.

### 01 Reserved GPU rental
- Committed capacity from verified neoclouds and hyperscalers.
- 1-36+ month terms Single normalized quote
- Verified neocloud partners

### 02 Token & forward contracts
- Future-dated GPU and inference-token delivery at pricing locked today.
- Future-dated delivery Lock in current pricing
- Bilateral contract terms

### 03 Open hardware market
- Used and refurbished data-center GPUs from verified sellers.
- Used + refurbished tiers 8 GPU classes
- covered Verified seller program

Keynote
Carmen Li / CEO compute.exchange

### 00:06:00–00:06:20

# Three ways to buy compute.

## Product lines

### 01 Reserved GPU rental
- Committed capacity from verified neoclouds and hyperscalers.
- 1-36+ month terms Single normalized quote
- Verified neocloud partners

### 02 Token & forward contracts
- Future-dated GPU and inference-token delivery at pricing locked today.
- Future-dated delivery Lock in current pricing
- Bilateral contract terms

### 03 Open hardware market
- Used and refurbished data-center GPUs from verified sellers.
- Used + refurbished tiers 8 GPU classes
- covered Verified seller program

---
Keynote
Carmen Li / CEO
compute.exchange

### 00:06:20–00:06:40

# Product lines
## Three ways to buy compute.

### 01 Reserved GPU rental
- Committed capacity from verified neoclouds and hyperscalers.
- 1-36+ month terms Single normalized quote
- Verified neocloud partners

### 02 Token & forward contracts
- Future-dated GPU and inference-token delivery at pricing locked today.
- Future-dated delivery Lock in current pricing
- Bilateral contract terms

### 03 Open hardware market
- Used and refurbished data-center GPUs from verified sellers.
- Used + refurbished tiers 8 GPU classes
- covered Verified seller program

### 00:06:40–00:07:00

### Product lines
# Three ways to buy compute.

### 01 Reserved GPU rental
- Committed capacity from verified neoclouds and hyperscalers.
- 1-36+ month terms Single normalized quote
- Verified neocloud partners

### 02 Token & forward contracts
- Future-dated GPU and inference-token delivery at pricing locked today.
- Future-dated delivery Lock in current pricing
- Bilateral contract terms

### 03 Open hardware market
- Used and refurbished data-center GPUs from verified sellers.
- Used + refurbished tiers 8 GPU classes
- covered Verified seller program

### Keynote
Carmen Li / CEO compute.exchange

### 00:07:00–00:07:20

### Product lines
# Three ways to buy compute.

## 01 Reserved GPU rental
- Committed capacity from verified neoclouds and hyperscalers.
- 1-36+ month terms Single normalized quote
- Verified neocloud partners

## 02 Token & forward contracts
- Future-dated GPU and inference-token delivery at pricing locked today.
- Future-dated delivery Lock in current pricing
- Bilateral contract terms

## 03 Open hardware market
- Used and refurbished data-center GPUs from verified sellers.
- Used + refurbished tiers 8 GPU classes
- covered Verified seller program

### 00:07:20–00:07:40

### Product lines
# Three ways to buy compute.

1.  **Reserved GPU rental**
    *   Committed capacity from verified neoclouds and hyperscalers.
    *   1-36+ month terms Single normalized quote
    *   Verified neocloud partners

2.  **Token & forward contracts**
    *   Future-dated GPU and inference-token delivery at pricing locked today.
    *   Future-dated delivery Lock in current pricing
    *   Bilateral contract terms

3.  **Open hardware market**
    *   Used and refurbished data-center GPUs from verified sellers.
    *   Used + refurbished tiers 8 GPU classes
    *   covered Verified seller program

### 00:07:40–00:08:00

### Product · 01 · Reserved GPU rental
# Reserved GPUs, without the rate card.

- **Guaranteed capacity**
  Contractual guarantee for the full window. Priority over on-

### 00:08:00–00:08:20

### Product · 01 · Reserved GPU rental
# Reserved GPUs, without the rate card.

- **Guaranteed capacity**
    - Contractual guarantee for the full window. Priority over on-demand traffic during shortage events.
- **Locked unit economics**
    - One rate for the term. Budget without exposure to on-demand volatility.
- **One normalized quote**
    - Compare hyperscalers, AI-native clouds, and specialty providers side by side in a single quote response.
- **Verified counterparties**
    - Identity, capacity, SLA, and operational checks before any quote surfaces.

### Committed terms
## 1–36+
month terms, quoted across the provider network

### Coverage
Current and next-gen NVIDIA platforms Bare metal or virtualized US · EU · APAC regions Hyperscalers, neoclouds, specialty

### Keynote
Carmen Li / CEO compute.exchange

### 00:08:20–00:08:40

# Segment live compute pools.

Portal • Marketplace

Every current and next generation pool with full specifications: memory, bandwidth, compute throughput, and interconnect. Configure or submit interest in one click.

### Screenshot of "Browse Compute Pools" application

[Screenshot of a web application displaying a list of compute pools with details like GPU model, memory, bandwidth, and throughput, alongside filtering options.]

### 00:08:40–00:09:00

# Requirement to RFQ in two minutes.

Portal - Request for quote

Pick vendor, model, and deployment type, then fabric, CPU platform, storage, and region. The form knows the cluster

### 00:09:00–00:09:20

# Set timing. Submit. We route it.

Lease preferences capture start date and term. On submit, our team reviews the request and selects the providers best suited to quote.

##

### 00:09:20–00:09:40

# Set timing. Submit. We route it.

Lease preferences capture start date and term. On submit, our team reviews the request and selects the providers best suited to quote.

### 00:09:40–00:10:00

# Set timing. Submit. We route it.

Lease preferences capture start date and term. On submit, our team reviews the request and selects the providers best suited to quote.

## New Request

### 00:10:00–00:10:20

### Product · 03 · Hardware market
# Own the silicon.

Pre-owned, refurbished, and OEM-surplus NVIDIA data center GPUs from identity-verified sellers. H10

### 00:10:20–00:10:40

### Platform - Inventory tracking
# Live inventory from every channel.

Provider inventory arrives via API, Slack, and email, and is normalized into one tracked book with GPU type, quantity,

### 00:10:40–00:11:00

### Platform · Operations
## Every RFQ, tracked end to end.
Full lifecycle across routing, quote review, and terminal states, with deal leads, provider assignments, and an audit trail on

### 00:11:00–00:11:20

## A verified network, not a listing site.
Corporate registration, beneficial ownership, operational track record, and signed partner agreements, all verified before a provider lists capacity.
[Screenshot of a dark-themed web application interface displaying a list of "Providers" with columns for partner type, organization, status, contact, and partner agreement status.]

### 00:11:20–00:11:40

### Platform - Quality scoring
# Providers spend time on serious demand.

Every buyer and every request is scored. Scores decide what reaches providers and how quickly.

### Buyer score
Built from account and activity signals, and updated as buyers submit and progress requests.

### Request score
Each quote request is scored on completeness and fit, including the buyer's lease preferences.

### Score gate
Only qualified buyers can express interest in provider listings.

Keynote
Carmen Li / CEO

### 00:11:40–00:12:00

# GPU Performance Data for Compute.Exchange Marketplace

[Screenshot of a web application displaying a marketplace for GPU listings, each showing GPU type, go-live date, duration, price, and location]

### 00:12:00–00:12:20

# AI Engineer NEW YORK
Presented by **arize**

## Keynote
Carmen Li / CEO compute.exchange

### Compute Exchange Request a Quote Form

[Screenshot of a web form

### 00:12:20–00:12:40

# NEW REQUEST FOR QUOTE
[Screenshot of a web application form for requesting a GPU quote]

### 00:12:40–00:13:00

# Keynote
## Silicon Index
Carmen Li / CEO compute.exchange

[Screenshot of a web application dashboard displaying GPU index prices and a line graph showing the price trend for GPU INDEX -

### 00:13:00–00:13:20

# AI Engineer NEW YORK
## Presented by Arize
### Keynote
Carmen Li / CEO compute.exchange

### Screenshot of Silicon Data: Silicon Index
*   GPU Index - H10

### 00:13:20–00:13:40

### Term-Implied Forward Curve
[Screenshot of a web application displaying a line chart of reserve term structure and implied forward prices over time, with associated data tables.]

### 00:13:40–00:14:00

# Term-Implied Forward Curve
[Screenshot of a web application interface displaying a line graph titled "Reserve Term Structure and Implied Forward Price - H100" and a data table below it.]

### 00:14:00–00:14:20

# Term-Implied Forward Curve
[Line chart showing Term-Implied Forward Rate over time for different dates]
[Table displaying Term Rate Curve data for various dates and maturities]

### 00:14:20–00:14:40

### Term-Implied Forward Curve
[Screenshot of a web application interface displaying a "Term-Implied Forward Curve" dashboard.]
[Line graph showing multiple forward rate curves over a 36

### 00:14:40–00:15:00

### Term-Implied Forward Curve
[Screenshot of the Silicon Data web application showing a line graph of Term Rate Curve and Implied Forward Rate over time, with a data table below detailing implied forward rates for different dates and contract durations.]

### 00:15:00–00:15:20

# Term-Implied Forward Curve
[Screenshot of a web application showing a line chart of implied forward rates over time for different dates, and a data table below it.]

### 00:15:20–00:15:40

### Term-Implied Forward Curve
[Screenshot of a web application showing a line graph of Reserve Term Structure and Implied Forward Price for B200 over time, and a table of Term Rate Curve data for various dates and timeframes.]

### 00:15:40–00:16:00

# Term-Implied Forward Curve
[Line graph showing multiple forward curves over 36 months, with values ranging from $4.0 to $6.4]
[Table displaying Term Rate Curve data for various dates, with values for 12, 24, and 36 months]

### 00:16:00–00:16:20

### Term-Implied Forward Curve
[Screenshot of a web application dashboard showing a line graph of Term-Implied Forward Curve values over time and a table of Term Rate Curve data.]

### 00:16:20–00:16:40

# NVIDIA AI Infrastructure Retains Value Beyond Accelerated Depreciation Schedules
8-GPU HGX System Cost Per GPU • Dotted Curve = Accelerated Schedule, 25% Residual at Year 2, Zero by Year 5

Residual value: Silicon Data, monthly averages. Purchase price basis: A100 $20,000 • H100 $35,000 • B200 $45,000. Mass availability dates assumed; purchase prices are third party estimates, not Silicon Data data.

[Line graph showing residual value of NVIDIA AI infrastructure (B200, H100, A100) over years since mass availability, compared to an accelerated depreciation schedule. The graph shows B200 at 158%, H100 at 58%, and A100 at 25% residual value, all significantly above the accelerated depreciation curve.]

### 00:16:40–00:17:00

# NVIDIA AI Infrastructure Retains Value Beyond Accelerated Depreciation Schedules
8-GPU HGX System Cost Per GPU • Dotted Curve = Accelerated Schedule, 25% Residual at Year 2, Zero by

### 00:17:00–00:17:20

# NVIDIA AI Infrastructure Retains Value Beyond Accelerated Depreciation Schedules

8-GPU HGX System Cost Per GPU • Dotted Curve = Accelerated Schedule, 25% Residual at Year 2,

### 00:17:20–00:17:40

# NVIDIA AI Infrastructure Retains Value Beyond Accelerated Depreciation Schedules

8-GPU HGX System Cost Per GPU • Dotted Curve = Accelerated Schedule, 25% Residual at Year 2, Zero

### 00:17:40–00:18:00

# NVIDIA AI Infrastructure Retains Value Beyond Accelerated Depreciation Schedules

8-GPU HGX System Cost Per GPU • Dotted Curve = Accelerated Schedule, 25% Residual at Year 2