Skip to content
FabriX

FabriX

The Global Compute Exchange.

FabriX is building a market and control plane for compute across providers.

Explore the schedulerSee the exchange

Working scheduler

Real decisions. Illustrative supply.

The scheduler filters, scores and allocates against the request you choose. The infrastructure dataset is a demonstration model.

fabrix://schedulerDemo · Ready
Allocated
0
Clearing
$0/hr
Latency
Pending
Carbon
Pending
Try

Reference network · illustrative data

Example capacity
18.4M
accelerators
Example regions
247
regions
Modelled utilisation
71.0%
of fleet
Modelled power
11.4 GW
reference network
Carbon objective
92.0%
of workloads
Example jobs
142,882
modelled
Modelled allocation
1.8s
to first node
01Exchange

A market for compute capacity.

The exchange will let providers publish available capacity and customers buy against a clear price. Spot, reserved and forward terms will share one market interface.

Explore Exchange
Illustrative market · USD per accelerator-hour
Order book14 example levels
Illustrative ask side of the order book, with capacity offered by providers
PriceUnitsRegionCpty
$2.45214,912Virginia12
$2.4236,528Frankfurt6
$2.4096,464Stockholm7
$2.3914,288Singapore6
$2.3731,664London3
$2.3602,752Virginia6
$2.3442,304Frankfurt8
Spread$2.336$0.008$2.344
Illustrative bid side of the order book, with capacity sought by customers
$2.3363,136London11
$2.3211,664Singapore4
$2.3006,976Stockholm12
$2.2875,824Frankfurt8
$2.26510,304Virginia11
$2.25110,240London10
$2.2446,208Singapore5
H200 · 48 hExample pricing
Last
$2.340
Session
−1.42%
Spread
38 bps
Open interest
486.2K
SpotReserved · 12 M48 h
Capacity calendarExample capacity by hour
Illustrative free accelerator capacity by region across the UTC day, as a percentage of each region’s total.
Region00:00 UTC01:00 UTC02:00 UTC03:00 UTC04:00 UTC05:00 UTC06:00 UTC07:00 UTC08:00 UTC09:00 UTC10:00 UTC11:00 UTC12:00 UTC13:00 UTC14:00 UTC15:00 UTC16:00 UTC17:00 UTC18:00 UTC19:00 UTC20:00 UTC21:00 UTC22:00 UTC23:00 UTC
VirginiaVirginia, 00:00 UTC: 19% freeVirginia, 01:00 UTC: 26% freeVirginia, 02:00 UTC: 31% freeVirginia, 03:00 UTC: 35% freeVirginia, 04:00 UTC: 38% freeVirginia, 05:00 UTC: 42% freeVirginia, 06:00 UTC: 44% freeVirginia, 07:00 UTC: 40% freeVirginia, 08:00 UTC: 47% freeVirginia, 09:00 UTC: 38% freeVirginia, 10:00 UTC: 37% freeVirginia, 11:00 UTC: 31% freeVirginia, 12:00 UTC: 33% freeVirginia, 13:00 UTC: 28% freeVirginia, 14:00 UTC: 19% freeVirginia, 15:00 UTC: 16% freeVirginia, 16:00 UTC: 9% freeVirginia, 17:00 UTC: 9% freeVirginia, 18:00 UTC: 10% freeVirginia, 19:00 UTC: 5% freeVirginia, 20:00 UTC: 6% freeVirginia, 21:00 UTC: 10% freeVirginia, 22:00 UTC: 9% freeVirginia, 23:00 UTC: 13% free
OregonOregon, 00:00 UTC: 22% freeOregon, 01:00 UTC: 19% freeOregon, 02:00 UTC: 23% freeOregon, 03:00 UTC: 30% freeOregon, 04:00 UTC: 32% freeOregon, 05:00 UTC: 39% freeOregon, 06:00 UTC: 44% freeOregon, 07:00 UTC: 48% freeOregon, 08:00 UTC: 51% freeOregon, 09:00 UTC: 51% freeOregon, 10:00 UTC: 53% freeOregon, 11:00 UTC: 52% freeOregon, 12:00 UTC: 48% freeOregon, 13:00 UTC: 46% freeOregon, 14:00 UTC: 41% freeOregon, 15:00 UTC: 42% freeOregon, 16:00 UTC: 34% freeOregon, 17:00 UTC: 31% freeOregon, 18:00 UTC: 23% freeOregon, 19:00 UTC: 26% freeOregon, 20:00 UTC: 21% freeOregon, 21:00 UTC: 20% freeOregon, 22:00 UTC: 16% freeOregon, 23:00 UTC: 20% free
LondonLondon, 00:00 UTC: 38% freeLondon, 01:00 UTC: 40% freeLondon, 02:00 UTC: 42% freeLondon, 03:00 UTC: 39% freeLondon, 04:00 UTC: 38% freeLondon, 05:00 UTC: 38% freeLondon, 06:00 UTC: 35% freeLondon, 07:00 UTC: 26% freeLondon, 08:00 UTC: 19% freeLondon, 09:00 UTC: 15% freeLondon, 10:00 UTC: 14% freeLondon, 11:00 UTC: 11% freeLondon, 12:00 UTC: 6% freeLondon, 13:00 UTC: 2% freeLondon, 14:00 UTC: 5% freeLondon, 15:00 UTC: 4% freeLondon, 16:00 UTC: 9% freeLondon, 17:00 UTC: 6% freeLondon, 18:00 UTC: 12% freeLondon, 19:00 UTC: 15% freeLondon, 20:00 UTC: 21% freeLondon, 21:00 UTC: 24% freeLondon, 22:00 UTC: 36% freeLondon, 23:00 UTC: 32% free
FrankfurtFrankfurt, 00:00 UTC: 45% freeFrankfurt, 01:00 UTC: 45% freeFrankfurt, 02:00 UTC: 49% freeFrankfurt, 03:00 UTC: 46% freeFrankfurt, 04:00 UTC: 42% freeFrankfurt, 05:00 UTC: 37% freeFrankfurt, 06:00 UTC: 34% freeFrankfurt, 07:00 UTC: 26% freeFrankfurt, 08:00 UTC: 27% freeFrankfurt, 09:00 UTC: 17% freeFrankfurt, 10:00 UTC: 13% freeFrankfurt, 11:00 UTC: 11% freeFrankfurt, 12:00 UTC: 12% freeFrankfurt, 13:00 UTC: 10% freeFrankfurt, 14:00 UTC: 9% freeFrankfurt, 15:00 UTC: 14% freeFrankfurt, 16:00 UTC: 15% freeFrankfurt, 17:00 UTC: 17% freeFrankfurt, 18:00 UTC: 22% freeFrankfurt, 19:00 UTC: 33% freeFrankfurt, 20:00 UTC: 32% freeFrankfurt, 21:00 UTC: 39% freeFrankfurt, 22:00 UTC: 45% freeFrankfurt, 23:00 UTC: 42% free
WarsawWarsaw, 00:00 UTC: 55% freeWarsaw, 01:00 UTC: 60% freeWarsaw, 02:00 UTC: 56% freeWarsaw, 03:00 UTC: 54% freeWarsaw, 04:00 UTC: 57% freeWarsaw, 05:00 UTC: 50% freeWarsaw, 06:00 UTC: 51% freeWarsaw, 07:00 UTC: 41% freeWarsaw, 08:00 UTC: 41% freeWarsaw, 09:00 UTC: 37% freeWarsaw, 10:00 UTC: 34% freeWarsaw, 11:00 UTC: 34% freeWarsaw, 12:00 UTC: 33% freeWarsaw, 13:00 UTC: 33% freeWarsaw, 14:00 UTC: 34% freeWarsaw, 15:00 UTC: 33% freeWarsaw, 16:00 UTC: 37% freeWarsaw, 17:00 UTC: 37% freeWarsaw, 18:00 UTC: 43% freeWarsaw, 19:00 UTC: 43% freeWarsaw, 20:00 UTC: 48% freeWarsaw, 21:00 UTC: 53% freeWarsaw, 22:00 UTC: 51% freeWarsaw, 23:00 UTC: 60% free
StockholmStockholm, 00:00 UTC: 62% freeStockholm, 01:00 UTC: 63% freeStockholm, 02:00 UTC: 60% freeStockholm, 03:00 UTC: 58% freeStockholm, 04:00 UTC: 60% freeStockholm, 05:00 UTC: 56% freeStockholm, 06:00 UTC: 47% freeStockholm, 07:00 UTC: 50% freeStockholm, 08:00 UTC: 41% freeStockholm, 09:00 UTC: 38% freeStockholm, 10:00 UTC: 34% freeStockholm, 11:00 UTC: 32% freeStockholm, 12:00 UTC: 36% freeStockholm, 13:00 UTC: 33% freeStockholm, 14:00 UTC: 38% freeStockholm, 15:00 UTC: 37% freeStockholm, 16:00 UTC: 39% freeStockholm, 17:00 UTC: 41% freeStockholm, 18:00 UTC: 45% freeStockholm, 19:00 UTC: 47% freeStockholm, 20:00 UTC: 47% freeStockholm, 21:00 UTC: 58% freeStockholm, 22:00 UTC: 55% freeStockholm, 23:00 UTC: 63% free
DubaiDubai, 00:00 UTC: 56% freeDubai, 01:00 UTC: 53% freeDubai, 02:00 UTC: 50% freeDubai, 03:00 UTC: 41% freeDubai, 04:00 UTC: 39% freeDubai, 05:00 UTC: 35% freeDubai, 06:00 UTC: 34% freeDubai, 07:00 UTC: 31% freeDubai, 08:00 UTC: 24% freeDubai, 09:00 UTC: 22% freeDubai, 10:00 UTC: 21% freeDubai, 11:00 UTC: 27% freeDubai, 12:00 UTC: 28% freeDubai, 13:00 UTC: 31% freeDubai, 14:00 UTC: 35% freeDubai, 15:00 UTC: 37% freeDubai, 16:00 UTC: 38% freeDubai, 17:00 UTC: 47% freeDubai, 18:00 UTC: 47% freeDubai, 19:00 UTC: 47% freeDubai, 20:00 UTC: 53% freeDubai, 21:00 UTC: 52% freeDubai, 22:00 UTC: 58% freeDubai, 23:00 UTC: 56% free
MumbaiMumbai, 00:00 UTC: 42% freeMumbai, 01:00 UTC: 33% freeMumbai, 02:00 UTC: 29% freeMumbai, 03:00 UTC: 29% freeMumbai, 04:00 UTC: 24% freeMumbai, 05:00 UTC: 15% freeMumbai, 06:00 UTC: 14% freeMumbai, 07:00 UTC: 8% freeMumbai, 08:00 UTC: 8% freeMumbai, 09:00 UTC: 7% freeMumbai, 10:00 UTC: 10% freeMumbai, 11:00 UTC: 10% freeMumbai, 12:00 UTC: 10% freeMumbai, 13:00 UTC: 18% freeMumbai, 14:00 UTC: 22% freeMumbai, 15:00 UTC: 27% freeMumbai, 16:00 UTC: 31% freeMumbai, 17:00 UTC: 36% freeMumbai, 18:00 UTC: 41% freeMumbai, 19:00 UTC: 44% freeMumbai, 20:00 UTC: 42% freeMumbai, 21:00 UTC: 45% freeMumbai, 22:00 UTC: 44% freeMumbai, 23:00 UTC: 39% free
SingaporeSingapore, 00:00 UTC: 19% freeSingapore, 01:00 UTC: 17% freeSingapore, 02:00 UTC: 6% freeSingapore, 03:00 UTC: 2% freeSingapore, 04:00 UTC: 2% freeSingapore, 05:00 UTC: 2% freeSingapore, 06:00 UTC: 2% freeSingapore, 07:00 UTC: 2% freeSingapore, 08:00 UTC: 2% freeSingapore, 09:00 UTC: 2% freeSingapore, 10:00 UTC: 2% freeSingapore, 11:00 UTC: 2% freeSingapore, 12:00 UTC: 11% freeSingapore, 13:00 UTC: 17% freeSingapore, 14:00 UTC: 17% freeSingapore, 15:00 UTC: 22% freeSingapore, 16:00 UTC: 31% freeSingapore, 17:00 UTC: 36% freeSingapore, 18:00 UTC: 38% freeSingapore, 19:00 UTC: 39% freeSingapore, 20:00 UTC: 37% freeSingapore, 21:00 UTC: 30% freeSingapore, 22:00 UTC: 32% freeSingapore, 23:00 UTC: 24% free
TokyoTokyo, 00:00 UTC: 29% freeTokyo, 01:00 UTC: 23% freeTokyo, 02:00 UTC: 18% freeTokyo, 03:00 UTC: 18% freeTokyo, 04:00 UTC: 14% freeTokyo, 05:00 UTC: 17% freeTokyo, 06:00 UTC: 13% freeTokyo, 07:00 UTC: 15% freeTokyo, 08:00 UTC: 19% freeTokyo, 09:00 UTC: 21% freeTokyo, 10:00 UTC: 25% freeTokyo, 11:00 UTC: 28% freeTokyo, 12:00 UTC: 33% freeTokyo, 13:00 UTC: 37% freeTokyo, 14:00 UTC: 43% freeTokyo, 15:00 UTC: 48% freeTokyo, 16:00 UTC: 49% freeTokyo, 17:00 UTC: 49% freeTokyo, 18:00 UTC: 44% freeTokyo, 19:00 UTC: 42% freeTokyo, 20:00 UTC: 41% freeTokyo, 21:00 UTC: 39% freeTokyo, 22:00 UTC: 37% freeTokyo, 23:00 UTC: 29% free
SydneySydney, 00:00 UTC: 34% freeSydney, 01:00 UTC: 35% freeSydney, 02:00 UTC: 31% freeSydney, 03:00 UTC: 29% freeSydney, 04:00 UTC: 25% freeSydney, 05:00 UTC: 29% freeSydney, 06:00 UTC: 31% freeSydney, 07:00 UTC: 31% freeSydney, 08:00 UTC: 33% freeSydney, 09:00 UTC: 42% freeSydney, 10:00 UTC: 40% freeSydney, 11:00 UTC: 45% freeSydney, 12:00 UTC: 50% freeSydney, 13:00 UTC: 52% freeSydney, 14:00 UTC: 56% freeSydney, 15:00 UTC: 58% freeSydney, 16:00 UTC: 54% freeSydney, 17:00 UTC: 56% freeSydney, 18:00 UTC: 53% freeSydney, 19:00 UTC: 51% freeSydney, 20:00 UTC: 53% freeSydney, 21:00 UTC: 43% freeSydney, 22:00 UTC: 39% freeSydney, 23:00 UTC: 41% free
São PauloSão Paulo, 00:00 UTC: 36% freeSão Paulo, 01:00 UTC: 40% freeSão Paulo, 02:00 UTC: 44% freeSão Paulo, 03:00 UTC: 43% freeSão Paulo, 04:00 UTC: 47% freeSão Paulo, 05:00 UTC: 51% freeSão Paulo, 06:00 UTC: 44% freeSão Paulo, 07:00 UTC: 42% freeSão Paulo, 08:00 UTC: 44% freeSão Paulo, 09:00 UTC: 38% freeSão Paulo, 10:00 UTC: 36% freeSão Paulo, 11:00 UTC: 29% freeSão Paulo, 12:00 UTC: 30% freeSão Paulo, 13:00 UTC: 24% freeSão Paulo, 14:00 UTC: 22% freeSão Paulo, 15:00 UTC: 19% freeSão Paulo, 16:00 UTC: 14% freeSão Paulo, 17:00 UTC: 14% freeSão Paulo, 18:00 UTC: 15% freeSão Paulo, 19:00 UTC: 18% freeSão Paulo, 20:00 UTC: 19% freeSão Paulo, 21:00 UTC: 22% freeSão Paulo, 22:00 UTC: 26% freeSão Paulo, 23:00 UTC: 32% free
Hour, UTC
TightFree
02Fabric

One control plane for compute.

Customers will describe the workload and its constraints. The fabric will choose eligible supply, coordinate placement and return one endpoint without exposing provider-specific infrastructure.

How the Fabric works
CustomerOne API. One contract. One bill.
  • Training
  • Inference
  • Batch
  • Agents
FabricThe intelligence layer above infrastructure.
  • Scheduler
  • Exchange
  • Router
  • Policy
  • Telemetry
  • Settlement
ProvidersProvider diversity behind one interface.
  • Hyperscale
  • Regional
  • Sovereign
  • Edge
03Intelligence

Scheduling is a policy decision.

Cost, latency, carbon, availability and residency are competing inputs. The policy model below uses real filtering and scoring logic against demonstration data. Change the objective to inspect the result.

Explore scheduling intelligence
Objective weightsPolicy demo

No single axis dominates. The scheduler trades cost against latency and carbon until the marginal gain on each is equal.

Cost
24%
Latency
24%
Carbon
20%
Availability
22%
Residency
10%
Candidate regions8 of 8 example regions eligible
  • Paris
    0.70
  • Stockholm
    0.69
  • Virginia
    0.67
  • Frankfurt
    0.61
  • Warsaw
    0.53
  • London
    0.49
  • Dublin
    0.36
  • Dubai
    0.36

Paris clears every constraint at once: 6 ms, $2.19/hr and 56 g/kWh with headroom to grow.

Paris is selected for the balanced objective. Paris clears every constraint at once: 6 ms, $2.19/hr and 56 g/kWh with headroom to grow.

CostLatencyCarbonAvailabilityResidency
04Energy

Route compute with the grid.

Power price and carbon intensity change by hour and region. FabriX is designed to make those signals part of placement, alongside latency, capacity and residency.

Explore energy-aware compute
Renewable-matched92%of modelled work
Median carbon128example gCO₂e per kWh
Energy arbitrage34%modelled cost reduction
06:00 UTC18 modelled flows · 58.6 GWExample placement → Stockholm
North Sea · wind · 12.4 GWHornsea · wind · 3.6 GWBaltic Array · wind · 4.2 GWNorwegian Fjords · hydro · 33 GWHengill · geothermal · 0.8 GWLoire Valley · nuclear · 61.4 GWIberian Plateau · solar · 9.8 GWOuarzazate · solar · 5.2 GWColumbia River · hydro · 21.6 GWLlano Estacado · wind · 18.2 GWBaie-James · hydro · 37 GWGreat Lakes · nuclear · 12.8 GWParaná Basin · hydro · 14 GWAtacama · solar · 6.4 GWAl Dhafra · solar · 4.8 GWAl Kharsaah · solar · 1.6 GWBhadla · solar · 8.4 GWGansu Corridor · wind · 20 GWKyushu · solar · 3.2 GWKori · nuclear · 24.6 GWSouth Australia · wind · 5.4 GWNorthern Cape · solar · 3.8 GWVirginia · 4% renewableOregon · 97% renewableToronto · 56% renewableSão Paulo · 58% renewableSantiago · 72% renewableLondon · 28% renewableFrankfurt · 44% renewableParis · 4% renewableWarsaw · 22% renewableDublin · 28% renewableStockholm · 80% renewableReykjavík · 69% renewableDubai · 41% renewableDoha · 38% renewableMumbai · 47% renewableSingapore · 48% renewableTokyo · 36% renewableSeoul · 4% renewableSydney · 49% renewableJohannesburg · 29% renewable
06:00 UTC
Lowest modelled cost · 06:00 UTC
  1. 1
    Stockholm
    80% renewable
    $24/MWh
    9 g/kWh
  2. 2
    Reykjavík
    69% renewable
    $25/MWh
    6 g/kWh
  3. 3
    Santiago
    72% renewable
    $29/MWh
    37 g/kWh
  4. 4
    Toronto
    56% renewable
    $30/MWh
    27 g/kWh
  5. 5
    São Paulo
    58% renewable
    $34/MWh
    61 g/kWh
  6. 6
    Oregon
    97% renewable
    $35/MWh
    19 g/kWh
windsolarhydrogeothermalnuclear
05Network

See capacity before you place it.

This explorer models how capacity, latency, power and residency could be inspected across providers. The topology and readings below are simulated.

Open Network Explorer
Primary regions20 example regions
Frankfurt
Frankfurt scenarioDE
Available5.0 ms$2.48/hr
Example capacity
246,000
Modelled utilisation
74%
Modelled power
610 MW
Carbon intensity
214 g/kWh
Power sources
WindSolarGrid mix
Accelerators (4)
GB200B200H200H100
Providers (2)
  • Northwind ComputeHyperscale99.96%
  • Halcyon DCRegional99.87%
Residency
GDPREU Sovereign
Backbone (5)
  • London11 ms · 96 Tb/s
  • Warsaw14 ms · 56 Tb/s
  • Paris9 ms · 72 Tb/s
  • Stockholm21 ms · 44 Tb/s
  • Dubai96 ms · 38 Tb/s
Region indexSelect to inspect
Example regions with illustrative capacity, utilisation, latency, power source and status.
RegionCapacityUtilisedLatencyStatus
512K78%6 msAvailable
288K69%11 msAvailable
96K61%14 msAvailable
64K72%34 msAvailable
28K88%41 msLimited
184K81%4 msAvailable
246K74%5 msAvailable
132K66%6 msAvailable
118K58%9 msAvailable
74K91%8 msLimited
92K55%12 msAvailable
46K49%19 msAvailable
156K63%16 msAvailable
88K51%18 msAvailable
128K77%22 msAvailable
142K89%7 msLimited
168K72%5 msAvailable
114K68%8 msAvailable
68K60%12 msAvailable
34K42%28 msOffline
07Developers

One request across every toolchain.

The proposed API describes accelerators, location, latency, residency and the scheduling objective once. The same shape can serve an SDK, CLI, Terraform provider or Kubernetes operator.

allocate.tstypescript · Concept preview
// Proposed API surfaceimport { FabriX } from "@fabrix/sdk"; const fabrix = new FabriX({ apiKey: process.env.FABRIX_API_KEY }); // Ask for capability. The fabric decides where it comes from.const allocation = await fabrix.compute.request({  accelerators: 4096,  type: "H200",  region: "eu",  maxLatencyMs: 15,  residency: ["eu-sovereign"],  optimise: "balanced",}); console.log(allocation.placements);// [ { region: "paris", units: 1536, price: 2.19 },//   { region: "frankfurt", units: 2048, price: 2.48 },//   { region: "stockholm", units: 512, price: 1.76 } ] await allocation.deploy({ image: "ghcr.io/acme/trainer:2.4.1" });
  • TypeScript
    TypeScript client
    Preview
  • Python
    Python client
    Preview
  • CLI
    Command line interface
    Preview
  • Terraform
    Infrastructure provider
    Preview
  • Kubernetes
    Cluster operator
    Preview
  • REST
    HTTP API
    Preview
08Enterprise

Controls belong in the architecture.

FabriX is being designed so residency, isolation, key custody and audit policy travel with the workload across providers. These are architecture goals, not current certifications or service-level claims.

Data residency

Placement policy will be able to exclude capacity outside an approved jurisdiction.

Encryption boundaries

The architecture treats encryption at rest, in transit and in use as placement requirements.

Customer-managed keys

The control plane is designed to integrate with customer-owned key management.

Tenant isolation

Isolation policy can require dedicated node pools and network segments.

Multi-region failover

Checkpoint and placement policy can prepare workloads to move after a provider incident.

Decision records

Placement inputs, provider selection and price are designed to produce exportable records.

Policy-bound
Placement

Constraints remain attached to every scheduling decision.

Portable
Recovery

Workloads are described independently of a single provider.

Traceable
Decisions

Placement inputs and outputs are designed to be exportable.

Help shape the compute exchange.

Request accessSee how it works

We are speaking with compute buyers, infrastructure operators and enterprise design partners.