How to afford AI

IT in Manufacturing

Engineering design software suppliers

Industrial software for PC suppliers

Manufacture/Management system suppliers

printer friendly version

September 2024 IT in Manufacturing

By Graham Brown, regional director for South Africa and SADC at Commvault.

Graham Brown, regional director for South Africa and SADC at Commvault.

Science fiction enthusiasts are familiar with AI, often depicted as the technology driving robots, cyborgs and spaceships with human-like thinking and computer-like efficiency. The business world recognises AI’s potential, and invests heavily in its development, seeing it as a key to technological advancement, competitiveness and profitability. Technology giants like Microsoft, Meta and Google emphasise AI’s significance.

The price tag of AI

AI can indeed improve decision making, efficiency, cost-effectiveness and customer relations. However, it comes at a substantial cost. Training models like ChatGPT and other Large Language Models (LLMs) demand significant financial resources. For instance, ChatGPT’s training used over 10 000 graphics processing units (GPUs), each costing around R567 223. Operating ChatGPT alone can cost up to R13 235 222 per day due to infrastructure requirements.

AI and cloud infrastructure

AI development goes beyond GPUs; it involves additional hardware and infrastructure, often relying on servers. Interestingly, these same servers used for data protection now democratise it. Cloud infrastructure plays a vital role in enabling LLMs, machine learning (ML) and the widespread use of automation. Reducing AI costs becomes crucial to business profitability.

Businesses can optimise their cloud-based operations by ensuring data integrity, while cutting costs. Leveraging Data Protection as a Service (DPaaS) is one such approach. DPaaS helps trim operational expenses related to cloud infrastructure, offsetting the rising AI development costs. This aligns with ongoing business optimisation efforts, fostering growth through cost reduction and reinvestment.

The future of data protection

AI-driven data protection offers continuous innovation to meet modern organisations’ evolving needs. Whether in the cloud or on premises, adapting to this evolving landscape is essential. Envision a future where you can automate tasks, analyse data in real time, achieve more with fewer resources, make informed decisions, receive precise security alerts, and utilise data insights for efficient customer and employee support.

These capabilities will soon become standard. Your data protection is the gateway to affordable AI adoption, offering endless possibilities.

Beyond current TCO concerns

When considering DPaaS, enterprises prioritise security, while assessing total cost of ownership (TCO). However, evaluating TCO can be challenging because of its quantitative nature and potential vendor bias. To navigate this effectively, consider these key factors:

• Deduplication and data tiering: Maximise storage efficiency and control cloud expenses.

• Cloud-native storage: Collaborate with DPaaS providers, for reduced storage costs, alongside your cloud service.

• Proactive ransomware monitoring and alerting: Ensure end-to-end data encryption and immutability for data security.

• Multi-cloud data protection: Choose DPaaS solutions aligned with diverse cloud needs.

• Support for various workloads: Your DPaaS solution should accommodate a mix of SaaS, IaaS, and on-premises resources.

• Robust reporting: Include compliance analytics to ensure adherence and future readiness.

Meeting these criteria will facilitate cost-effective data protection, enabling you to plan and reinvest your savings wisely.

Share this article:

IT in Manufacturing

How to afford AI

Further reading:

Published by Technews