Empowering Indian Enterprises with SpeedCloud AI: Democratizing Access to Cost-Effective and Tailored AI Solutions

CXOToday has engaged in an exclusive interview with Mr. A S Rajgopal, CEO & Managing Director of NxtGen

1. What inspired NxtGen to develop SpeedCloud AI, and how does it address the specific needs of Indian businesses regarding AI adoption? Can you elaborate on the industries you plan to target with these specialized AI tools
   1. SpeedCloud^® is an indigenous cloud service, that delivers up to 80% cost savings compared to global cloud providers and delivers superior performance, thanks to opensource technologies, high-density compute, All NVMe storage and high-bandwidth, low latency networks. The principle was to ensure fair pricing for our customers. We are working on the same principles on the SpeedCloud^® AI platform as well.
   2. Enterprises cannot use GPUs, it is important that we solve real world business problems for our customers, our focus will be deliver a comprehensive platform with repository of models from NVIDIA, AMD & Intel, software for customising or developing use-cases, generic use-cases and we are also going to host a lot of use-cases developed by the start-up community on SaaS model for our enterprise consumption.
   3. Generic use-cases available for free to use on https://speedcloud.nxtgen.com are sovereign to the user. These are intended for enterprises to experiment and understand what to expect from AI.
      1. We will be releasing our model, Accura this month to support Chartered Accountants and enterprise finance departments. We are also in advanced stage of training Juris to support corporate legal teams and subsequently add more laws to support lawyers in their day-to-day productivity.
      2. We have delivered two use-cases to Govt of India establishments and working on fraud detection for financial services and some models for healthcare sector.

2. How does NxtGen plan to democratize access to cloud and AI technologies for enterprises,particularly smaller businesses or startups that may not have the resources to invest in thesecloud services?

Small businesses have access to AI Compute resources, provided it is viable for their business. We see a large market for inference ie. Running use-cases, rather than training very large models, it is not possible to match the opensource options being rolled out globally.

1. We are providing free-to-use Sovereign to user AI use-cases, small businesses can confidently use these generic use-cases without fear of losing their intellectual property. The AI models are sophisticated based on Llama 3.1/3.3, IBM and other industry leading options, that run on high-end GPU infrastructure. The intent is to introduce AI into the small businesses, so that they can imagine how they leverage AI for specific use-cases for their businesses.
   1. Employee productivity can be enhanced with use of Aurora.
   2. Use ADA to build new applications rapidly for free.
   3. Vizia & Lumina models help visualize or extract information from visuals.
   4. Accura will help addressing finance queries and Juris corporate law.
   5. We should be able to introduce one model every 3 months free to use.
2. Leveraging AI for a business is not just about GPUs, you need experts to help build use-cases. We are offering services to curate and ingest data, model selection & fine-tuning and built-in guard rails ensure security of the AI model.
   1. We use open-source to keep AI open and eliminate cost of proprietary models.
   2. Being platform agnostic helps us keep the hardware costs lower, by offering optimised solutions across NVIDIA, AMD & Intel platforms.
   3. We are encouraging many startups to build on our platform and helping them to take their solutions to our customers.

Democratising AI is about making AI work for small businesses at cost that makes sense for their business.

3. Can you share some examples of successful AI use cases that NxtGen has enabled for its customers, and what impact have these had on their businesses?
   1. Our customers are seeing immediate (results in few minutes) using our developer model ADA, we are helping our customer to understanding security & stability issues and to modernize code for their existing applications. Reduce up to 40% time in coding new applications. We now have over 2,000 developers from customers across verticals using this tool.
   2. We have also trained a custom model for a constitution body (GoI) to provide information on laws, processes, rules etc., which we expect to be rolled out for 50+ lakh officials.
   3. The free to use models now based on Llama 3.3 are being used by over 10,000 users and it helps understand how to optimize GPU resources for large scale deployments.

4. Can you elaborate on the partnership with Akash Systems and how Diamond Cooling technology will enhance NxtGen’s AI computing services? What kind of performance improvements can customers expect?
   1. For AI to be successful, managing cost is one of the most important tasks, the use-cases should make business sense in terms of cost at which they are delivered.
   2. Working with Akash to deploy world’s most energy efficient GPU servers. Akash is introducing a layer of diamond on the Chips to extract heat 5 times faster than current copper heat-sinks and use water-less liquid cooling system to dissipate heat. With this we expect to achieve 1.151 PUE compared to 1.6 – 1.7 PUE, resulting in lower power consumption for managing heat for GPUs. Further this technology eliminates high-speed fans, typically there are 18 of these in a GPU server consuming 30watts each.
   3. These servers are expected to consumer 40% less energy and these cost savings will allow us to be globally competitive with a fair pricing model for our customers.

5. How does NxtGen plan to expand its infrastructure and services to meet the growing demand for data center services in India, and what kind of investments is the company making in this regard?
   1. Our current infrastructure is at its peak of utilization. Our existing customer base has 140 Peta Bytes of production data, we are bringing sovereign AI to their data.
   2. We are investing in most recent hardware – AMD MI325X and NVIDIA H200SXM platforms. We expect 1024 of these to be delivered in February and available for consumption in March, about Rs.300 crores investment. At this point we see a consumption of 512 GPUs every quarter, about Rs.150Crores.
   3. We expect to be empaneled by INDIAai to provide AI. Compute resources, this would require us to double our infrastructure immediately. We have commitments from NVIDIA and AMD to ensure that we have access to these quantities.

6. How does SpeedCloud AI unify GPU platforms from NVIDIA, AMD, and Intel, and what advantages does this provide to enterprises building AI use cases?
   1. NVIDIA is a winner today by a big margin, however, we have to ensure our customers have an optimal choice of platforms to run their use-cases more efficiently.
   2. We worked with Red Hat OpenShift to unify all the platforms, so that our customers have a single interface and a choice of hardware below to run their use-cases.
   3. My team is fortunate to learn from experts at NVIDIA, AMD and Intel, we believe this learning along with opensource technologies will ensure cost-effective solutions for our customers.

7. What kind of cost savings can customers expect with NxtGen’s energy-efficient AI compute services, and how will this impact their bottom line?
   1. We will ensure 60 – 70% cost savings on AI compute for our customers compared to popularly available choices – a combination of cost saving and fair-pricing the services. We have already published our pricing on our web-site.
   2. Today most use-cases with GenAI are focused on employee productivity. GenAI can deliver up to 95% accuracy, which when compared humans on a vast knowledge is significantly more and the models work tirelessly 24/7, I believe virtual assistants to help deliver superior customer service are the next to widely deployed.
   3. Quantification of impact to bottom-line depends on maturity of the deployment by the customer. In one use-case, cost of training 50+ lakh officials for 3-days on-site vs providing a ready to use AI model for reference, the cost savings are up to 90% and it delivers improved quality and eliminates errors in the functioning of the officials.

The post Empowering Indian Enterprises with SpeedCloud AI: Democratizing Access to Cost-Effective and Tailored AI Solutions appeared first on CXOToday.com.