Elastic AI & Machine Learning as a Service Platform-Proposal

Telcos/Service Provider can play major role in transformation of Singapore to be a major AI hub & help in job creation

I took a example Singapore and did market research. It covers business idea, customer research, market research, trend analysis, customer strength, platform design & flow, platform setup & operating cost, revenue model, managed services.

Overview

Elastic AI & Machine learning on-demand platform which will help Startups, Education institution, Enterprises, Government sector in faster Adoption of AI & Machine Learning in Singapore & APAC region

Problems to solve

1. Traditional GPU is Less efficient
2. Need large space & power
3. Longer time for go to market
4. Hardware is costly
5. High Operating & Setup Cost
6. skill shortage, data & use case what to do with data

Project objective

To provide AI & Machine Learning on-demand platform which will accelerate the AI/Machine Learning/ adoption in Singapore & APAC region.

Understanding the market

Asia has highest concentration of smart cities

60% of world population is concentrated in Asia

Asia Pacific Machine learning market is anticipated to exhibit CAGR 32.92% over the forecast period of 2019–2027

Singapore has advantage of Infra access & talent pool which telcos/service providers can tap to promote the adoption

AI will not increase the unemployment, it will transform the existing jobs to new

Market trends

AI Index — Singapore

Reference: Ref: http://vibrancy.aiindex.org/

Targeted Sector

1.| Healthcare

2.| Public Sector for Smart Cities

3.| Agriculture

4.| Finance

5.| Robotics

Proposed solution

1. Digital Marketplace for AI/ML as a service
2. Integrated Machine Learning Stack
3. Remote vGPU
4. Automated on-demand service
5. Network for GPU over the network

Advantages of platform for customer

1. Easy access to GPU & Machine Learning platform

2. Faster go to market

3. Cheap & Less operating cost

4. Data security as data will be on-premise while vGPU will be attached over network locally

5. Datacenter footprint reduction

6. Startups will get easy & cheap access to AI/ML Infra

7. Service Providers/Telcos can collaborate with NUS, NTU & local universities

8. It will help in Job creation

Process Flow

Total Operating Cost

Cost Comparison

Ref: https://determined.ai/blog/cloud-v-onprem/

Approx Cost: 1200 $ / VM / Month

Avg Cost / Customer : 12,000 $ / Month (Assumption 10 instances)

Avg Cost / 100 customers: 1.2 M $ / Month

Revenue Stream

1.Machine Learning Stack

2.Customisation of stack / customer requirement

3.GPUaaS — Usage Based

4.GPUaaS — Subscription based

5.GPUaaS — Dedicated

6.Support

7.Managed Services

8.Training

9.Network charges for vGPU access

Summary