No Exclusives
Arvind Krishna (Chairman and CEO at IBM), Shantanu Narayen (Chair and CEO at Adobe), Punit Renjen (Global CEO Emeritus at Deloitte and incoming SAP chairman) and Brad Smith (Vice-chair and president at Microsoft) – these are the giants of tech leadership that I had missed out on for exclusive interviews in recent times as the very pantheon of technology world descended to grace the B20 Summit in New Delhi in late August.
As a journalist, I thrive and breathe on exclusive stories. And there is nothing bigger than missing out on the biggest stories featuring the biggest names by the slimmest of margins. So, when I got wind that Jensen Huang, Founder, President and CEO at Nvidia, was in India, I was the first to pitch for an interview – only to be denied.
Instead, I received an invitation to a "closed-door meeting" with select members of the media, offering a chance to listen to Huang's confidential conference in Bengaluru. My disappointment knew no bounds, even though I heard that the opportunity was limited and journalists from Mumbai and Delhi who were keen on attending this exclusive briefing were left disgruntled.
During a discreet and hurried briefing prior to the meeting, the chosen journalists were presented with strict protocols: just one question per journalist, a baffling ban on recording (which was later allowed), a prohibition on phone usage (supposedly to avoid disturbing Huang) and a strict no-photography policy. The invitation itself stipulated, "Please reach the NVIDIA office by 2:30 pm at the latest for security check. There will be no entry allowed post 2:30 pm" – an unconventional directive. Interestingly, the meeting began well after 3 pm with a hard stop set at 4 pm.
The Leather Jacket CEO
As Huang made his entrance, clad in his trademark leather jacket, at Nvidia's Mahadevapura (Bengaluru) office, the room's chatter abruptly hushed. Yet, Huang appeared visibly uneasy as he noticed people moving about beyond the translucent glass walls of the conference room. Without hesitation, he stepped outside and summoned those who were lingering, impatiently inquiring about the arrival of certain other individuals.
He made a pointed reference to us, the journalists, jesting that these latecomers would be assessing our behaviour. Despite this, his restlessness remained evident as he urged Nvidia staff to take their seats, all the while showing eagerness to commence, contingent upon the mysterious latecomers' arrival. It struck me that he possessed the traits of a stern leader, a notion corroborated by a former employee who had worked directly under Huang and currently held a C-level position at a data cloud company.
"The most important people are the employees. You guys know that, right? Let them in. Let them in," implored the Nvidia CEO, as he invited those standing outside the conference room to join our gathering.
"Nvidia employees are the most valuable assets in our company. Our company's priorities are crystal clear. We take care of our employees and we take care of our families. In return, they naturally produce exceptional products, hopefully driving substantial sales that benefit our shareholders," he added with a touch of humour.
Nevertheless, the meeting commenced without the anticipated late arrivals. Huang began by reminiscing, "It's been five years since I've been here (India). Five years ago, I spent time with Modiji (Indian Prime Minister) because he invited me to talk. I was fortunate to have addressed Modiji's cabinet and spoke to him about artificial intelligence, specifically focusing on this revolutionary concept of software authored by computers, with the computer engineer embedded within the machine."
Huang was referring to generative AI (GenAI), a groundbreaking technology that had captured the world's attention since the public release of OpenAI's ChatGPT in late November 2022. This technology has instilled both fear and excitement among people. But across the business landscape, companies worldwide are scrambling to devise strategies to harness this cutting-edge technology.
Indeed, according to a Goldman Sachs analysis, GenAI has the potential to catalyse an impressive 7 per cent surge in global GDP, equating to nearly $7 trillion and spur productivity growth by 1.5 percentage points over the span of a decade. Following a comprehensive examination of databases encompassing task profiles for over 900 occupations, Goldman Sachs economists projected that approximately two-thirds of jobs in the United States could be influenced to some degree by AI-driven automation. This projection takes on even greater significance when extended beyond the confines of the United States, hinting at profound global ramifications in the years ahead.
But Huang confidently expressed his lack of concern regarding job displacement. "The most widely used application for AI today is software programming. Currently, around 60-70 per cent of the code on GitHub is generated by AI. Yet, not a single software engineer has lost their job," he asserted.
Huang attributed this to the shared aspiration to enhance software development, doing so more efficiently and effectively. "Nobody lost their jobs. We simply invested in more AI. Thus, the only change we have witnessed is in the curriculum of computer science courses," Huang concluded.
India & AI
There is no question that the US, with companies such as OpenAI, Google, Microsoft, Meta and IBM, has the lead on GenAI and AI as a whole among all countries. Second, of course, is China. Chinese organisations launched 79 large-language models (LLMs) in the country over the past three years as they doubled down on efforts to develop artificial intelligence (AI) algorithms, according to a report prepared
by Chinese-run research institutes. Meanwhile, India, which quintessentially drives IT services for the globe's 200,000 enterprises, lags on this front.
Huang said India needs to work on its AI infrastructure rapidly for the development of LLMs and AI proliferation. He stated, "Artificial intelligence, just like 5G, needs infrastructure." He went on and announced two ambitious AI infrastructure projects: one with Reliance and another with the Tata Group.
"We are going to build AI supercomputers with Reliance. They are going to use it to create generative AI services for their 450 million customers," Huang revealed.
"In the case of Tata, they are going to create AI infrastructure. This infrastructure will support all of the AI companies in India, which are clamouring for infrastructure. They are flying all over the world looking for infrastructure. They might have to export their engineers to go work in California so they can have infrastructure there. There's no reason to do that. Build the infrastructure here. The engineers don't have to leave," Huang said.
Interestingly, a Bain & Company report says that India is the third-largest contributor to global Artificial Intelligence (AI) talent. The country is responsible for 16 per cent of global AI talent despite constituting a small share of the global AI market. The report surmises that this significant share of AI talent is due to the technology workforce of the country growing up in an internet/cloud-first world.
The study also says that the micro, small and medium enterprises (MSMEs) in India are punching above their weight by leveraging their workforce and specialist contractors to incorporate AI/ML modules into their ecommerce ventures and digital business processes. The report also observed that India has seen a huge spike in AI talent demand from large MNCs looking to transform their operations through AI/ML after the pandemic.
Under the terms of the partnership with Tata, Nvidia said it will also help upskill over 6 lakh employees at TCS on AI.
GPUs Coming To India
One of the journalists asked about how many GPUs India would be sourcing to grow fast in AI domain as the Reliance and Tata deals with Nvidia bear fruits – given OpenAI’s five-year journey to reach their current position. In response, Huang estimated that India would require approximately tens of thousands of GPUs to establish the necessary infrastructure, specifically around 1,00,000 GPUs or fewer, equating to fewer than 400 AI supercomputers. Huang expressed optimism about India's AI prospects and predicted that India could achieve results comparable to those of US companies like OpenAI in a matter of weeks rather than months when it comes to training GPT models.
Why and how? Huang asserted that Nvidia is preparing to introduce the world's fastest computers, which have not yet been produced. India will be among the first countries to access these computers, and they are expected to surpass anything previously seen in terms of speed while remaining cost-effective. He anticipated that by the end of the next year, India will possess AI supercomputers that are at least 50 to 100 times faster, significantly reducing the cost associated with training foundational AI models.
In the initial stages, both Reliance and Tata will have access to DGX Cloud, a service that enables any business to utilise its AI supercomputer via a standard web browser, simplifying the process of acquiring, deploying and managing on-premises infrastructure. Huang highlighted that it is now feasible for anyone and everyone to construct foundational AI models, such as ChatGPT, for a cost ranging from $10 million to $20 million, thanks to Nvidia’s efforts to make it more affordable by leveraging existing cloud infrastructure.
On my part, I revisited a query I had posed to Huang during an event last year, inquiring about India's prospects in the realm of AI. Back then, he had emphasised that India's true potential lay in companies like TCS, Infosys and similar entities, which could propel AI into enterprises worldwide. This time, I tweaked my question, shifting the focus to generative AI, given that behemoths like Google, OpenAI and Microsoft appeared to dominate the market. I asked, 'What opportunities does India have in this context?'
Huang responded by emphasising that in our highly specialised world, no single entity, whether a corporation or an individual, could possess all the world's knowledge or intelligence. He acknowledged the notion of a potential AI singularity in the future but underscored the importance of practicality in the present. He argued for harnessing existing technology and training AI systems to excel in specific domains, such as accounting, law and chemistry, to effectively address various needs.
Amused that Huang broached the topic of the 'AI singularity,' I couldn't resist asking the inevitable question: 'Are you concerned about sentient AI?' His succinct response was 'No.' He chose to conclude the discussion by remarking, 'There are countless other considerations to ponder over.'
Leaving the 'closed-door meeting,' albeit one that was not exclusive, I departed with more insights to share. As a tech journalist, that's ultimately what matters most.
