GPU cloud for financial services
GPU capacity for banks, insurers and asset managers, with location, provider and ownership shown for every offer.
Context
Financial institutions use AI for risk and fraud models, document processing and assistants for advisers and operations. Training and running these models needs GPU capacity that most institutions do not own.
Using an external cloud counts as relying on an ICT third-party provider. Supervisors expect institutions to know where data is processed, who the provider is and how they can exit.
Rules
Overview of key rules, starting with Europe and Switzerland. Not legal advice. Have your obligations reviewed by legal counsel.
DORA, Regulation (EU) 2022/2554
Applies since 17 January 2025 to financial entities in the EU: ICT risk management and oversight of ICT third-party providers, including a register of contractual arrangements.
Source: eur-lex.europa.euFINMA Circular 2018/3 “Outsourcing”
Requirements for Swiss banks and insurers that outsource significant functions, including access rights and data location.
Source: www.finma.chGDPR and Swiss FADP
Personal data of clients may only be processed by providers that offer sufficient guarantees, under a data processing agreement.
Source: eur-lex.europa.eu
What Intelaris solves
Documented location
Every offer shows country and city. Filter by the country you need.
Known provider and owner
Seat of the provider and of its parent company for every offer.
Comparable prices
Public prices with source and date for budgets and internal approvals.
One request
One request to several providers instead of one per cloud.
Typical workloads in Financial services
Which GPU fits depends on the work, not on the industry. Shown are GPUs with current offers whose manufacturer names the use case.
Inference
Running a trained model to answer requests, for example a chatbot or an image check.
GPUs with current offers
- H100
- A100
- L40S
- L4
- H200
- RTX PRO 6000 Blackwell
- V100
- B300
- B200
- A30
- L40
- Instinct MI300X
- GB200
- T4
- RTX 4000 Ada
- A40
- RTX A6000
- RTX 6000 Ada
- A10
- Gaudi 3
- GB300
Fine-tuning
Adapting an existing model to your own data. Needs far less compute than training from scratch.
GPUs with current offers
All GPUs for this use caseHPC and simulation
Scientific computing and simulation, for example weather, materials or flow models.
GPUs with current offers
All GPUs for this use case
Prices
Current prices per GPU-hour, with source and date.
| GPU | Index, USDEURCHF | Cheapest current offer, USDEURCHF | Providers |
|---|---|---|---|
| H100 prices | 3.753.353.09 | 2.642.362.18 | 28 |
| H200 prices | 5.645.044.65 | 3.363.002.77 | 21 |
| L40S prices | 1.511.341.24 | 1.201.070.99 | 16 |
As of Oct 25, 2026, 21:05 UTCSource: public price lists and pricing APIs of the providersConverted at ECB reference rates of Oct 23, 2026Method
Financial services
Offers from several providers for your GPU, quantity, term and location.