Certainly! Having been actively engaged in the AI research and engineering field since its formative years, I’m happy to share some insights on how the AI industry is evolving. I hope this sheds light on the topic.
AI is a computation-intensive domain that demands high-end GPUs to execute its tasks effectively. Currently, the industry relies heavily on two major manufacturers: Nvidia and AMD. Companies like E2E serve as outsourcing partners, leveraging collaborations (e.g., E2E’s Gold Sponsorship for the Nvidia AI Summit 2024 in Mumbai) to provide infrastructure solutions to businesses. These partnerships enable firms from startups to large-scale enterprises like Jio, L&T, and other major Indian companies to access cutting-edge AI technology.
Also, E2E is also actively collaborating (or seeking to) with premier institutions in India to accelerate AI research and innovation, creating a robust ecosystem for AI development.
As for ROI, AI investments are yielding impressive returns. Companies are successfully pitching AI-powered products to VCs and securing substantial funding. While it’s true that some AI startups face challenges and fail, the broader trend shows continued financial commitment from corporations, governments, and academic institutions to advance AI capabilities.
To sum up, the AI industry is thriving, with E2E emerging as a leader in building infrastructure for companies, governments, and educational institutions. AI continues to shape the future, driven by significant investments and innovations.
As per earlier intimation, LnT gets 2 board seats.
Ms. Shrimati Ambastha
is the Chief Executive
of Data Center
Business of L & T and
holds Master’s degree
in computer science
with over 33 years of
extensive experience in IT, data center
management, and cloud
technologies. She held
key leadership
positions at Oracle and
VMware for more than
15 years, followed by
her role as SEVP at
NTT India GDC for
over 7 Years.
“Infrastructure major Larsen & Toubro (L&T) on Wednesday (December 4) said it has successfully completed the acquisition of a 15% stake in E2E Networks Ltd through a preferential allotment. In addition to this, L&T is set to acquire a further 6% stake from the promoters of E2E Networks, with the secondary acquisition expected to be finalised on or before May 30, 2025.” → from cnbc article
So for first 15 percent shareholding percentages were diluted and then promoter gets to sell their stake directly?
So in theory when this was announced overnight migh stake in company was reduced by 15%. And now i do not get the chance to sell my shares to L&T or any other instituion.
Don’t really like how this deal happend. Maybe L&T will take it to new heights but such astronomical growth requires great money but on expense of minority share holders is not really a good look.
If the promoter sells in the market, the promoter gets the money whereas if the company does preferential allotment, the company gets the money.
Selling stake to L&T is also a strategic decision to make them eligible for bidding for government contracts and also improving their financial strength and sustainability.
The promoter has sold 6% stake at a huge discount @ 2750/- or so to LnT while the market price was hovering around 4500/–5000/- . I am not sure why any of us want to have a similar deal.
Yotta planning for 1 lakhs gpu in next 2 to 3:years…
Elon musk xAI is planning for 1 million GPUs from current 1 lakhs…
Based on this trend, I am hopeful of E2E has huge runway going forward, even if they able to deploy 10K GPUs, that’s 3k cr capex. Which translates to 1200cr ebidta… ofcourse, it’s all depends on how AI theme play out…
News Update - Jio Platforms Partners with Nvidia to Revolutionize AI Space
So the real test of how good the tools that E2E has built, will begin now. Just like the ability of NVIDIA to offer a complete package in the form of the GPUs + CUDA is what gave them the competitive advantage, I hope E2E would be able to hold its own in the face of fresh competition, because if push comes to shove, they won’t be able to complete with Jio on price.
There is clear differentiating here on what Jio and E2E partnered with Nvidia.
Jio and TATA have partnered to develop language models.
Please enlighten if anything is missing, I am learning here.
That may be true but if I’ve learnt anything watching Reliance do anything, they don’t target niches. They takeover. They’re almost definitely working towards something larger. They define and create industries. How E2E fits into it, I have no idea.
It also said this "Joining the India AI Mission, Jio secures Nvidia’s Blackwell GPUs to deliver GPU-as-a-service at globally competitive rates. "
Now GPU-as-a-service, which is basically IaaS with GPUs offered for AI related computing (training, inference, et al) is at the core of E2E’s growth story. Reliance is pretty much the Amazon of India in terms of sweeping away the competition with their predatory pricing, and coffers large enough to outlast any competition.
companies must install a minimum of 1000 AI compute units
1 compute unit means 1 GPU?
And from where do these 1000 units come from? Does the govt (under this mission) provide to the empanelled companies the GPUs directly or provide funds to buy the GPUs? if the company already has some GPUs, will they get counted in these 1000?
Well, the foremost advantage that I find with E2E Networks Ltd and alike companies in India is the location cost advantage. In order to host the servers/GPUs they need space, and the cost of space is much less in India compared to countries outside like USA etc. Due to this, they can play with the price and offer cloud services at a lower cost which MS, Amazon, Google will never be able to provide unless they take a hit on their margins and I don’t think these biggies will ever try to experiment with it since the market is too big for all who are currently present and each being a leader in their domain.
E2E Networks Ltd provides the infra as a service at a very low cost which the other providers do not provide. If I have a small to medium business, I’ll not need very critical & complex integrations for which Paas, SaaS play an important role. For storage purpose and not so critical integrations I’ll just need a low-cost solution which does my job. Google, Amazon, Azure do not provide very low-cost solution for the same. This is where E2E Networks Ltd shines. If I grow bigger in size, I’ll need to have critical and complex integrations for which I can opt for AWS, Azure etc, for their PaaS solutions incurring higher costs and that’s pretty fine. The need for both Infra and Platform will be there, and I can opt for different CSPs respectively for each use case under single umbrella. This market is too big for all and having said that let’s not forget the GPU space separately as well. At some point I don’t doubt companies like JIO providing the GPU processing lending solutions at lower costs that no one can imagine. This particular market is so large that multiple CSPs can coexist together for a long time. TAM is very big here.
E2E Networks Ltd: L&T’s stake highligts big ambitions but does not provide immediate movement up the value chain. It excels in offering cost-effective, basic IaaS solutions for SMEs.
Jio: With potential GPU processing services and deep pocket , Jio could disrupt the market with low costs.
Yotta: Competes with E2E in India but caters to larger enterprises and a has much larger scale than E2E.
International Players (AWS, Google, Oracle): Offer comprehensive services but at higher costs, targeting larger and global clients. They are also find Indian Market attractive and may ties up with Indian partners for brining business.
Alibaba Cloud: Emerging as a strong contender in Asian markets due to cost advantages.
Overall TAM is huge and Growth rate nearly 20 % for next 5 years provide sufficient room to grow for all players. Being domestic provider is an additional advantage for E2E, it is recently shortlisted by govt for 10K Cr AI and cloud mission.
A critical review of the actual E2E network platform portal.
Disc: Invested.
I had signed up with E2E networks last year but didn’t really use their services apart from just clicking around on the portal. Today I signed up with intention of trying to use their server for a product we are building currently hosted on DigitalOcean. This is not a tech forum so I will keep the tech points brief:
First resource creation is filled with roadblocks. I was just trying to get a basic 4 vCPU server started which is a breeze on other services like DO, Hetzner, Linode. But on E2E it took me filling the form 4 times.
Even for a basic server (2,263 Rs/month), I had to stop with the creation process and pay 100 Rs. for account verification, 200+ Rs. for IP reserve and 84.50 Rs. for something else. This meant 3 times I had to pick up my phone and verify the transaction before I got to see an actual server.
The steps to fill the form was error-ing out with “something went wrong” pop-ups 2-3 times requiring me to start filling the form again
I accidentally clicked on another SSH key key after attaching the first and could not undo it, apart from refreshing the page and restarting the form from the beginning.
The VPC creation took substantially long and I actually had to wait for it to be created in the background to complete the server form. That’s just poor UX. Why not take my preference and build the VPC along with the server?
This is just immature testing
I have significant investment in the company and really wanted to use the server to test things out and I actually felt like giving up through the process several times. It feels like going to your kid’s song competition and feel like walking out while your kid sings cause they are that bad. I have close to 1.5 decades building tech stuff but this is outright the worst first-impression I’ve had with any service(including shared hosting we used in college days lol).
On the plus side,
The new account creation was smooth.
I got a call from E2E within 100 seconds of account creation.
Maybe the person could help but I don’t see how the road bumps would go away by having them on call
I might have had 1 of a million user flows which was severely glitched but there’s no way around so many micro transactions happening.
As a developer I am used to seeing resource creations fly through, fast creations and not be obstructed with “credit” purchase request while creating a new resource. Definitely self correcting and recoverable forms.
Suggestions:
The credits purchased hardly last 24 hours, why not just allow the resource to run without/free credit for the first day.
If some resources are too expensive (GPUs), ask for a prepay for those but allow some cheap resources to be created more easily.
If postpaid is not allowed by the business model, give free credits on signup, enough to run some small stuff for 24-48 hours.
If none of this is possible, charge a higher one time infra credit purchase at signup so at least those who pay have to pay once while trying things out and not be asked for repeated prompts.
Better UX, much more testing. I filled one form and almost gave up. UI Testing is taking a backseat trading off new features it seems.
This is not a rant post to get E2E to solve my case. I have significant investment in the company and no tech investment yet. I’d like to see them fix it.
Hopefully someone @E2E reads this, confirms it internally and allots resources to improve.
Very Good insight and lot of scope for improvement; hence revenue for E2E.
E2E operates in GPU based clouds with focus on cost and optimization, Although It has TIR platform, for developing AI models but lacks much behind Sagemaker ( AWS) and AI platform of google clouds in terms of end to end ML pipelines and advanced AI. Can its collobration with L&T, it may develop the capability to match with thoese of AWS, Google etc
E2E mainly operates in open source TensorFlow/PyTorch framework for traiming and building models while AWS, Amazon etc use proprietry, full stack ML/ DL process and are highly customized. E2E has narrower range of tools.
E2E has lesser pre-trained modelswhile AWS, Google etc has very large pre-trained models. Google even uses specialized accelarator TPU.However, E2E offers simple solutions for startup/ SMEs needs.
AWS/ google etc provide vast integration with their other services.
So, for at present E2E is better for local customized needs but limited AI capability. The question is :would they go up in value curve like of AWS/ google with advanced AI tools, integrating their services with other providers and expand globally
