How do you move your AI workloads to the cloud?

CloudMachine Learning (ML)
2.3k viewscircle icon5 Upvotescircle icon5 Comments
Sort by:
Director of IT in Healthcare and Biotech3 years ago

moving workload to cloud gives you cost efficiency, all the resources required,  newest technologies, and flexibility with business needs. AWS is my preference, while other player also offerf good value for money.

Lightbulb on1
Senior Product Marketer in Software3 years ago

AI workloads can indeed benefit from the cloud's scalability. If you move them to the cloud, you'll probably watch your compute costs skyrocket. This is true whether you choose AWS or Azure, and whether you deploy containers in Kubernetes or good-old VMs. 
So it's crucial to accompany this transformation with a serious cost optimizer that supports stateful workloads. AWS and Azure do have their native tooling for that, but 3rd party solutions usually generate way more savings. Just be sure to carefully inquire how their stateful support is delivered, and whether it satisfies your workloads' needs: Persist data, persist network, both, or something beyond that. Good luck. 

Global Vice President of Sales in Software4 years ago

AWS is the simplest and most cost effective platform - I would encourage you look into that platform.

C-Suite in Energy and Utilities4 years ago

First of all, it needs to be aligned to your cloud strategy, where Cloud native platforms will be critical for your AI workloads.

Lightbulb on2
Chief Information Officer in Finance (non-banking)7 years ago

AI workloads are best suited for cloud and the journey should actually begin from Cloud native platforms like tensorflow, AWS sagemaker, Azure ML etc. These platforms are built to handle massive data workloads and can expedite the training cycles for ML.

Lightbulb on2

Content you might like

Have you experimented with any new team structures or compositions today that wouldn’t have been previously possible without AI?

Which pitfalls—model bias, false positives/negatives, data quality, regulatory constraints—often impede AI-based security tools, and how can they be mitigated in a financial-services context?

Read More Comments

How long does your organization retain original systems logs used to filter SOX-related actions into a system that requires review of the logs and retains the filtered logs for seven years? Does your organization consider those original system logs records subject to record retention requirements, or supporting information used to create the SOX records?

90 Days14%

365 Days50%

3 years27%

5 years5%

7 years5%

Other (share in the comments)

View Results

Are you worried about shadow AI? Do you have a formalized governance process around it?

Read More Comments

How would you prioritize adding native vector search capabilities directly to core MySQL storage engines (e.g., InnoDB)?

Note: This poll is specifically about vector search as a general-purpose feature in the core database, separate from specialized analytics engines like MySQL HeatWave.

High Priority: This is a critical need. We want to run AI/vector workloads on our primary transactional data without relying on a separate database or a specialized analytics add-on.

Medium Priority: This would be a valuable feature for future projects. It would allow us to innovate on our core database, but it's not an immediate requirement.70%

Low Priority: This is a nice-to-have, but not essential. Our primary focus for MySQL remains on its traditional OLTP performance and stability.10%

Not a Priority: This is not a good fit for MySQL. We believe vector workloads are fundamentally different and are best handled by a dedicated system, keeping our core MySQL lean.20%

Unsure / Need More Information: We are not yet clear on the performance, security, or operational impact of integrating this capability into our core transactional engine.

View Results