Lenovo Brings ‘Made in India’ GPUs to Life

Lenovo has finally entered India and is building ‘Made in India’ GPUs and AI servers. The company inaugurated its SOTA research and development lab in Bengaluru, making it the fourth city to host an infrastructure R&D lab globally, after Raleigh, Beijing, and Taipei. The company operates 18 R&D labs in all.

Lenovo goes all-in 

“We spent over $2 billion on R&D, a growth of almost 50% year on year. One in every four employees are in our research and development [team], and the reason I'm calling that out is that it’s the core of the company,” said Amit Luthra, managing director, ISG, Lenovo India, at the launch of the AI Lab at Lenovo’s Bengaluru campus. 

The company also revealed its AI server manufacturing plans, announcing that AI-enabled GPU servers will be built right here in India. As many as 50,000 enterprise AI servers with 2,400 high-end GPU units will power local and global markets. 

“Today, the government has committed a 10,000+ GPU-strong ecosystem in public-private partnership to ensure that these AI workloads get the horsepower [needed] to run these technologies. Now with this, it's very important to see how Lenovo is actually grabbing this opportunity, and most importantly, helping organisations as they move ahead in this AI transformation journey,” said Luthra. 

India for the World

“I think it's a great opportunity for us to actually build the lab over here. It's no secret that there's a lot of innovation and great resources in India,” said Sumir Bhatia, president, Asia Pacific, ISG at Lenovo, in an exclusive interaction with AIM on the sidelines of the lab launch. 

Bhatia said that the lab isn’t just for India – it has a global focus. He believes it will be part of a larger billion-dollar project and will work with other labs that specialise in different areas, also ensuring that this approach allows all labs to collaborate and contribute to a global network.        

The collaboration paves the way for further partnerships as well. “We are very globally tied-in with the other design centres. We complement some of the areas… and it works out beautifully,” Satish Pratapneni, director, system dev at Lenovo, told AIM. 

He emphasised that the company does not take a one-sided approach. When it comes to full-fledged capabilities, they believe they have a lot to offer on the other side as well. He also highlighted why India's growing technical expertise is taking centre stage. 

Inside Lenovo’s new R&D lab in Bengaluru

Smarter AI for All 

The company has been going strong on its focus to build an end-to-end solution for AI infrastructure, from servers to PCs. It not only posted strong revenues in the recent quarterly report, but also attributed 47% of it to non-PC revenue.

The AI server unit in Puducherry will be Lenovo’s fifth manufacturing hub in the world, following USA, Mexico, Brazil, and Hungary. 

All this growth is also intertwined with a number of strategic partnerships. Top AI player, NVIDIA, has been a key collaborator for Lenovo. In October of last year, the two companies revealed their partnership and expansion plans aimed at making generative AI more accessible to enterprises. 

Bhatia stressed that partnering with industry giants runs deep in their ecosystem. “It’s in our DNA to make sure that we’re partnering with the best, because this is what will bring solutions to our customers. So, whether it’s NVIDIA, Red Hat, Nutanix, Intel, or AMD, we’ve got the whole ecosystem and are building platforms and solutions around that,” he said.

Lenovo has over 60 AI solution providers globally, and has delivered more than 165 solutions.         

Meanwhile, the company is focusing on building skilled resources from the ground up and investing in academia too. Discussions are underway with schools, including second and first-year programs. The goal is to engage more with institutions to train the trainers and embed the desired skills from the start. 

AI specialists is another area of investment. “We’ve got over 400 AI specialists who know AI end-to-end,” said Bhatia.

Enjoy the full story here. 

OpenAI o1 Likely Uses RL over CoT to Build System 2 LLMs

Subbarao Kambhampati, a professor at the Arizona State University, said that OpenAI’s o1 model uses reinforcement learning over auto-generated chain of thought—similar to AlphaGo’s self-play approach—to optimise problem-solving by building a generalised System 2 component atop LLM substrates, albeit without guarantees. 

“One interesting issue with o1 is that it seems to be significantly less steerable compared to LLMs. For example, it often completely ignores any output formatting instructions making it hard to automatically check its solutions,” he added. He further said that once you are an approximate reasoner, you might develop the ‘don’t tell me how to solve the problem; I already have a way I use to solve it’ complex.  

Know more about OpenAI o1 here. 

Agentforce Takes Center Stage at Dreamforce

This year, Dreamforce has been all about Agentforce. AIM is present at ground zero bringing live updates from the event, being held at Moscone Center in San Francisco. 

Salesforce launched over 100 pre-built industry-specific AI actions for Agentforce, allowing businesses in sectors like healthcare, retail, and manufacturing to quickly deploy tailored AI agents for customer success.

It has fully integrated Tableau into the Salesforce platform to help users access data visualisation, insights, and actions through Agentforce, making data-driven decisions faster and seamless.

Slack channels now integrate seamlessly with Agentforce, allowing human employees and AI agents to collaborate on workflows, share insights, and take actions directly within Slack.

Also, Salesforce revealed the upcoming version 2 of Agentforce, codenamed Atlas, which will deliver a 90-95% resolution rate on all services and sales cases, setting a new standard for AI performance in business. 

Other key highlights: 

• Atlas Reasoning Engine: The new Atlas Reasoning Engine powers Agent Force, utilising advanced AI techniques like RAG to enable agents to learn, plan, and take action across all Salesforce clouds.
• Salesforce Foundations: Salesforce announced Salesforce Foundations, a free upgrade for Enterprise Edition and higher, providing automatic access to Sales Cloud, Marketing Cloud, Service Cloud, Commerce Cloud and others. 
• Unified Salesforce Platform: All Salesforce apps, including Sales, Service, Marketing, and Commerce, are now rewritten on a unified platform to seamlessly integrate AI and automation, enhancing business productivity across the Customer 360 ecosystem.
• Salesforce Omni Supervisor: The new Omni Supervisor tool enables businesses to manage both human and AI agents in real time, providing visibility into all customer interactions across channels for optimal service management.
• Tableau Pulse for Real-Time Insights: Tableau Pulse, now GA, provides real-time business insights powered by AI from Data Cloud, allowing organisations to visualise, analyse, and act on customer data directly from within Salesforce.
• AI Trust Layer for Salesforce: Salesforce announced its AI Trust Layer, ensuring that all AI interactions are secure, trusted, and compliant with industry regulations, providing businesses with confidence in deploying AI agents.