The Innovation Beneath AI: The US-India Partnership powering the AI Era

20 Feb 2026 17:00h - 18:00h

Session at a glance

Summary

This panel discussion focused on the infrastructure requirements needed to support AI at scale, examining the physical foundations beneath AI development rather than just the models themselves. The conversation was moderated by Ujwal Kumar and featured experts from venture capital, entrepreneurship, government, and technology sectors discussing the massive infrastructure buildout required for AI deployment.

The panelists emphasized that AI’s scaling depends heavily on physical infrastructure including energy systems, semiconductors, critical minerals, and data centers. Tuan Ho from X Fund highlighted the strategic vulnerability created by supply chain dependencies, particularly in rare earth magnets essential for everything from hard drives to chip manufacturing. He noted that over 90% of rare earth magnets currently come through China, creating significant risks for AI infrastructure development.

Jeff Binder discussed how AI tools are creating unprecedented opportunities for entrepreneurs, allowing them to bring products to market with significantly less capital than previously required. However, he warned of potential overbuilding in the infrastructure space, similar to the fiber buildout during the early internet boom. Prince Dhawan from REC Limited explained India’s innovative approach through the India Energy Stack, which enables programmable power and peer-to-peer energy trading, allowing data centers to source power dynamically from distributed solar installations.

Vrushali Gaud from Google outlined the company’s $15 billion commitment to India, including new subsea cables and AI hubs, emphasizing India’s potential for leapfrogging traditional infrastructure limitations. Dr. Tobias Helbig from NXP Semiconductors provided a contrasting perspective, suggesting that the current focus on centralized data centers might be analogous to IBM’s prediction of only five computers worldwide, with the real future lying in billions of edge devices requiring minimal power.

The discussion concluded with agreement that while current infrastructure investments are necessary, the ultimate transformation will likely involve decentralized, edge-based AI systems that operate more efficiently and sustainably than today’s power-hungry data centers.

Keypoints

Major Discussion Points:

– AI Infrastructure Requirements and Scale: The panel extensively discussed how AI deployment at scale requires massive infrastructure buildout including energy systems, semiconductors, critical minerals, and data centers. This represents what Jensen Huang called “the largest infrastructure build-out in human history.”

– US-India Strategic Partnership in Critical Supply Chains: Significant focus on the US-India critical minerals corridor, rare earth supply chains, and Google’s $15 billion commitment to India including subsea cables and AI hubs. The discussion emphasized reducing dependence on China for critical materials like rare earth magnets.

– Energy Grid Transformation and Clean Power: Detailed exploration of how AI’s massive energy demands require “programmable power” and intelligent grids. India’s Energy Stack was highlighted as enabling peer-to-peer energy trading and coordination at scale to support data centers through distributed renewable sources.

– Evolution from Centralized to Edge AI Computing: The panel discussed a fundamental shift from power-hungry centralized data centers to efficient edge devices, comparing it to the evolution from IBM’s “five computers” to billions of personal devices. This transition could dramatically reduce power requirements while bringing AI closer to users.

– Investment Opportunities and Risks in AI Infrastructure: Discussion of where venture capital should focus, with emphasis on durable infrastructure investments over volatile GPU/model investments. The panel noted potential overbuilding risks similar to the dot-com era, but with better measurability of success/failure.

Overall Purpose:

The discussion aimed to shift focus from AI models and applications to the foundational infrastructure required to support AI at scale, exploring investment opportunities, policy frameworks, and technological innovations needed across the US-India partnership.

Overall Tone:

The tone was consistently optimistic and forward-looking throughout, with panelists expressing excitement about opportunities while acknowledging real challenges. There was strong consensus on the transformative potential of AI infrastructure, though some cautionary notes were raised about potential overbuilding and obsolescence risks. The discussion maintained a collaborative, solution-oriented atmosphere with panelists building on each other’s insights rather than disagreeing.

Speakers

Speakers from the provided list:

– Ujjwal Kumar – Founder and CEO of Quantum Alliance, Co-founder of Cognosy AI (Moderator)

– Tuan Ho – Unicorn founder turned venture capitalist, Partner at X Fund

– Jeff Binder – Serial entrepreneur with multiple Fortune 500 exits, Harvard Venture Partners

– Vrushali Gaud – Global Director of Climate Operations at Google

– Prince Dhawan – IAS officer, Executive Director at REC Limited under the Ministry of Power

– Tobias Helbig – VP of Innovation at NXP Semiconductors

– Participant – (Role/title not specified – appears to be event organizer/host introducing the panel)

Additional speakers:

None identified beyond the provided speakers names list.

Full session report

This comprehensive panel discussion examined the critical infrastructure requirements needed to support artificial intelligence deployment at scale, moving beyond the typical focus on AI models to explore the foundational physical systems that will enable the technology’s widespread adoption. Moderated by Ujwal Kumar, founder and CEO of Quantum Alliance and co-founder of Cognosy AI, the conversation brought together experts from venture capital, entrepreneurship, government policy, corporate strategy, and semiconductor innovation to address what Jensen Huang has called “the largest infrastructure build-out in human history.”

The Infrastructure Imperative

The panel opened with Kumar’s observation that whilst AI models receive significant attention, the underlying infrastructure requirements represent where the real opportunities and challenges lie. This infrastructure spans critical minerals, energy systems, semiconductors, data centres, and grid modernisation—all of which require unprecedented coordination and investment. Recent developments underscore this urgency, including Google’s $15 billion commitment to India featuring a gigawatt-scale AI hub in Vizag and new subsea cables connecting through Africa and Singapore-Australia routes.

Critical Supply Chain Vulnerabilities and Investment Opportunities

Tuan Ho, a unicorn founder turned venture capitalist at X Fund, provided crucial insights into the strategic vulnerabilities created by current supply chain dependencies, particularly in rare earth magnets. His firm’s investment in Vulcan Elements—now backed by $1.4 billion in funding—illustrates the scale of intervention required to address America’s rare earth magnet supply chain, where over 90% currently flows through China. This dependency creates profound risks because magnets are essential for virtually all moving components: hard drives, motors, and critically, the manufacturing of semiconductors themselves.

Ho emphasised that this represents a fundamental shift in venture capital focus, with investors now examining decades-old industries that have seen minimal innovation. Power grids that haven’t been upgraded for nearly a century, refining capacity limitations, and sustainable data centre operations all present significant opportunities. The convergence of AI infrastructure needs with geopolitical supply chain concerns creates what he described as “huge opportunity for us as investors” in building companies that address these foundational requirements.

Importantly, Ho noted that infrastructure businesses offer “more durability and clarity” regarding problem definition compared to pure AI model companies, as the underlying problems—power distribution, cooling, materials processing—remain consistent even as specific technologies evolve.

Entrepreneurial Leverage and Market Dynamics

Jeff Binder, a serial entrepreneur with multiple Fortune 500 exits now at Harvard Venture Partners, provided insights into how AI tools are transforming entrepreneurship itself. He argued that current AI capabilities give entrepreneurs “massive leverage” they previously lacked, particularly in cross-border collaboration between the US and India. AI tools are eliminating traditional barriers in front-end development, enabling entrepreneurs to deliver products with potentially “a tenth the capital” previously required.

However, Binder introduced a crucial contrarian perspective, projecting that within two years, the industry might face significant overcapacity and ROI challenges despite current concerns about power and compute shortages. This potential overbuild, paradoxically, could benefit entrepreneurs by making infrastructure resources extremely inexpensive. Importantly, he distinguished the current AI boom from the dot-com era by highlighting that “almost every aspect of artificial intelligence deployment from the foundational aspects all the way to the top of the stack are measurable,” making success and failure clearer and faster to determine.

India’s Strategic Energy Innovation

Prince Dhawan from REC Limited provided perhaps the most technically sophisticated analysis of AI’s energy requirements, introducing the concept that “AI essentially will not scale unless your power is programmable.” His central thesis identified intelligent grids, rather than chips or compute capacity, as the primary constraint for AI scaling.

Dhawan detailed India’s groundbreaking approach through the India Energy Stack, which creates interoperable digital rails enabling coordination at scale. The system allows data centres to source power dynamically from distributed sources, fundamentally changing the economics of AI infrastructure. As he explained, “individual retail households can essentially monetize their rooftop solar power by supplying to such data centers,” creating new livelihood opportunities through peer-to-peer energy trading.

The technical innovation lies in the stack’s ability to handle “standard rules for measurement, identification, and settlement all in near real-time,” enabling what Dhawan termed “intelligent electrons.” This addresses the fundamental challenge that whilst AI evolves in quarters, traditional grid infrastructure evolves over decades. Dhawan also highlighted Reliance’s commitment of “trillion dollars in the next seven years” as indicative of the scale of investment flowing into Indian infrastructure.

Google’s Full-Stack Infrastructure Strategy and Climate Innovation

Vrushali Gaud from Google provided insights into the company’s comprehensive approach to AI infrastructure, emphasising innovation across the entire technology stack. Google’s $15 billion commitment to India encompasses not only data centres but also networking infrastructure, including subsea cables creating global connectivity through multiple routes.

Gaud explained Google’s rationale for heavy investment in India beyond the obvious market size. India’s young, tech-eager population has demonstrated remarkable ability to leapfrog traditional development stages, as evidenced by rapid adoption of digital payments through UPI and GPay. This leapfrogging potential extends to clean energy infrastructure, where India represents one of the few places where “the math on clean energy just works.”

A significant announcement was Google’s Climate Technology Center, developed in partnership with the Office of Principal Scientific Advisory for the Government of India. This initiative focuses on three key areas: green skilling for decarbonisation careers, low-carbon materials for construction (including data centres), and sustainable aviation fuel development. Importantly, the center targets Tier 2 and Tier 3 cities and involves a “wider spread of universities” to democratise innovation opportunities and ensure contextual relevance to local conditions.

The Edge Computing Revolution

Dr. Tobias Helbig from NXP Semiconductors provided the most provocative perspective, using a historical analogy to challenge current assumptions about AI infrastructure. He recalled that in 1942, IBM’s head predicted a world market for “about five computers”—technically correct for that era’s computers, but completely missing the evolution toward billions of personal devices.

Helbig suggested that current discussions about massive, power-hungry data centres might represent a similar blind spot. He pointed to efficiency contrasts between human brains (20 watts) and flies (below 1 milliwatt for remarkable intelligence) to argue that dramatic improvements are possible. His company already demonstrates this potential: “we today have products where on whatever 10 watts or so you can run very meaningful LLMs,” enabling AI deployment at the edge rather than in centralised facilities.

This vision encompasses AI evolution “moving from, hey, I can perceive something, is it a dog, a cat, to I can think, generative AI, I can create something out of those models to the point that I can create agents” that act independently in the real world. His marathon watch, which runs for 12 days on a single charge whilst providing significant intelligence, exemplifies this future direction.

Helbig also offered perspective on innovation cycles, noting “we have a tendency to overestimate the next two years and impact and underestimate what’s happening in 10 years,” suggesting that edge computing advances may be more transformative than currently anticipated.

Investment Risk Assessment and Financial Innovation

The discussion revealed sophisticated understanding of investment risks across different infrastructure layers. Prince Dhawan noted that financial institutions already recognise obsolescence risks by refusing debt financing for GPUs whilst providing it for basic infrastructure like buildings and power systems, validating concerns about the durability of different infrastructure components.

However, Jeff Binder warned that hardware breakthroughs could potentially make entire data centres “almost instantly, at least from a financing perspective, obsolete” if new chip designs achieve dramatic efficiency improvements. This creates a complex risk landscape where investors must balance infrastructure durability against rapid technological advancement.

Unprecedented Government Financing

The panel highlighted unprecedented levels of government investment in AI infrastructure globally. Tuan Ho noted that having tech conferences where prime ministers and heads of state announce hundreds of billions in infrastructure investment represents something “we’re not used to seeing” from a venture capital perspective. This government financing spans multiple nations, creating global competition for AI infrastructure leadership whilst providing both opportunities and challenges for private investors.

US-India Partnership and Strategic Collaboration

Throughout the discussion, the US-India partnership emerged as a particularly promising model for international cooperation in AI infrastructure development. This collaboration combines India’s innovation potential, favourable clean energy economics, and digital infrastructure capabilities with US investment and technological expertise. The partnership spans venture capital investment, corporate strategy, government policy coordination, and technical innovation across the entire AI infrastructure stack.

Future Scenarios and Unresolved Challenges

The panel identified potential future scenarios ranging from successful coordination between distributed energy resources, intelligent grids, and efficient edge computing, to significant overbuilding of centralised infrastructure that becomes stranded as efficiency improvements reduce power requirements.

Several critical challenges remain unaddressed, including permitting issues for clean energy infrastructure, the mismatch between AI development timelines (quarters) and infrastructure development timelines (decades), and the need for financing models that appropriately price obsolescence risks across different infrastructure layers.

Conclusion

This panel discussion successfully shifted focus from AI models to foundational infrastructure requirements, revealing a complex landscape of opportunities, risks, and interdependencies. The consensus emerged that whilst current infrastructure investments are necessary, the ultimate transformation will likely involve more distributed, efficient systems operating closer to users and applications.

The discussion’s most valuable contribution was demonstrating that AI infrastructure requires simultaneous consideration of technical, financial, geopolitical, and environmental dimensions rather than optimising any single factor in isolation. As Tuan Ho observed, the unprecedented scale of government investment in AI infrastructure represents a new paradigm for venture capitalists and entrepreneurs alike. This holistic perspective, combined with the measurability advantages that distinguish AI deployment from previous technology waves, will be essential as the world navigates what may indeed prove to be the largest infrastructure buildout in human history.

Session transcript

Participant

Thank you. Thank you. Thank you. this infrastructure right now and closing the gap between commitments and capacity. This is where the real opportunity lives. Moderating today’s session is Ujwal Kumar, founder and CEO of Quantum Alliance and co -founder of Cognosy AI. Quantum Alliance works with universities, industry and governments to get top talent working on the foundational problems beneath AI, from critical minerals to energy to semiconductors. He will be joined by Tuan Ho, Unicorn founder turned venture capitalist, now partner at X Fund. Jeff Binder, serial entrepreneur with multiple Fortune 500 exits, now at Harvard Venture Partners. Prince Thavan, IAS officer and executive director at REC Limited under the Ministry of Power. Rushali Gaut, global director of climate operations at Google.

Dr. Tobias Helbig, V.I. and VP of Innovation at NXP Semiconductors. Ujwal, over to you. We’ll start with a quick picture for the panelists if you can all rise Thank you

Ujjwal Kumar

Thank you everyone We are up against Jan, we are up against her boss. So, but, let’s have fun in this panel. And the broader idea, like we have been hearing all about AI models, what AI can do, and this panel is more about, we are talking about AI at scale now, what it needs, what it would make fulfill when we talk about AI, like AI -driven companies, when we are talking about AI -driven solutions. Let’s talk about this now, as AI is forcing creative destruction of how the world builds infrastructure, energy, semiconductors, critical minerals, physical edge systems, data center. US and India are now building this together, rare earth corridors in India’s union budget. Google committed $15 billion to India and accelerated focus on clean energy.

Jensen at Davos called this the largest infrastructure build -out in human history. Two weeks ago, 54 countries launched FORGE, the first global framework for the minerals that power AI. Yesterday, on this very summit, Sundar Pichai laid out Google’s $15 billion commitment to India, a gigawatt -scale AI hub in Vizag, four new subsea cables between US and India. The models are getting attention, the infrastructure is getting the money, and we exactly have the right people to figure out where is all this going and what do we need further. Thank you. To start with, XFund was the early investor in Vulcan Elements. Now it is backed by 1 .4 billion US dollar government partnership to bring America’s rare earth magnet supply chain.

What, according to you, the US -India critical minerals corridor look like from the investor side?

Tuan Ho

First off, thank you for having us here. I’m really glad you pulled this panel together. I think your points earlier about the focus that we tend to put on discussing AI models and everything in the model layer, but we don’t really talk about what exists underneath that, is actually a really unique topic to cover and one that I and XFund, has generally been extremely excited about. so the way I look at that is that in this strive that we have to build intelligence we tend to talk a lot about the industrial revolution that it will create we often have to understand we often look at the industrial revolution that will be required in order to support the creation of that infrastructure there are a lot of inputs required for AI infrastructure so you’ve got energy you’ve got energy the power grid, power generation has to be clean, sustainable, renewable and the demands for AI infrastructure are going to require us to really solve large problems as to how to supply that power you’ve got, everybody’s talking about critical minerals now you mentioned Vulcan elements right You know, Vulcan Elements was a business that we invested in.

It was a Navy veteran out of Harvard who had spent a lot of time looking at supply chain issues for the U.S. military and noted that, you know, 95, over 90 % of, you know, magnets, rare earth magnets were coming through China. It creates a strategic vulnerability for the United States. And the reason why it creates a vulnerability is because, if you think about it, there are so many things that we need, that we build that require magnets. You can’t build hard drives. You can’t build motors. You can’t, I mean, nothing that moves can be built without them. We talk a lot about chips. You can’t manufacture chips without magnets. And so I look at, you know, problems like that, and for the first time, you know, I’m not going to be able to build a magnet.

I think you’ve got guys like me, venture capitalists, looking at… the opportunity to invest in building up that type of infrastructure to solve those sorts of problems. But that’s just one, right? You have to figure out how do you source it? Where do you get the materials from? And so when you look at things happening on the geopolitical scale, for the first time we are, at least in the United States, we’re looking at these trade deals to try to figure out where we’re going to supply the materials. Like how are we going to completely rebuild our power grid? How are we going to build up the capacity for refining those materials? Where are we going to source them from?

Where are we going to have to get them from? As we’re looking at data centers, how can we make them more sustainable, more power efficient? In order to support the AI needs that we have right now, like power consumption for data infrastructure, infrastructure is already, I think it’s about… It’s approaching 10 % plus. How are you going to meet that demand? And so from an investor perspective, yeah, we’re going to look at all of the cool AI products that entrepreneurs are looking to build. But on the other side, what is very exciting for us is looking at all the low -hanging fruit that exists for all of the inputs of industries that have not been innovated in for decades.

Power grids that have not been upgraded for the better part of a century. It creates huge opportunity for us as investors. And you mentioned US -India. Yeah. I… I find a lot of opportunity in the U.S. and India working more closely together to try to figure out how on both sides of the world we can build great companies to meet that need.

Ujjwal Kumar

Thanks, John. You spoke about needs. You spoke about the innovation. You spoke about what the early stage startups should be focusing on. I’ll move to Jeff, who has built companies from scratch and made multiple Fortune 500 exits. I would ask him what would it need for young entrepreneurs to build and scale in this space and do it successfully.

Jeff Binder

Thank you for having us and putting this together. I know that we have more people here than Sundar has at his keynote. So I heard Sam Altman only had 10 yesterday, so we’ve already outdone him. So, you know, I think it’s such an interesting time. I was there in the early web days in 99 and 2000 and 2001, and, you know, the excitement around the Internet obviously fueled a massive tech boom. And ultimately a fiber build -out, an infrastructure build -out, and it took years for all of that infrastructure to be absorbed, and ultimately it was. I think the difference this time around is that the tool sets themselves that entrepreneurs have available to them are smart.

And they can bridge some of the challenges, especially since we’re talking about partnerships between the U.S. and India. But, you know, oftentimes, especially when you get to things like user interfaces, there are cultural differences from the development work that would happen in India for an Indian audience or the U.S. or China. That’s always made it more difficult for collaboration, sort of on the front end. And many of the products that entrepreneurs are working on are often, you know, front end facing, consumer facing, at least initially. They’re generally not building a lot of B2B platforms. That happens later when you get the experience as to what’s necessary in a business environment. And I think that AI is going to change drastically the ability to leverage sort of cross -border talent, in particular with India and China and other places that were harder to leverage before.

It’s certainly from a quality perspective, SQA and back -end development, I think entrepreneurs have been able to leverage India and other places for the last couple decades. But it’s been harder to get the front end of a product to sort of match the cultural necessity of a given market. And I think that’s going to change. And I think for entrepreneurs, it means that they have a massive amount of leverage that they didn’t have before. And it means that we’re going to have a flood of new ideas that are actually brought to market and work fairly well and allow entrepreneurs to deliver products with probably a tenth the capital, depending on the product, obviously. If you’re doing magnets, you’re sort of stuck with the physical properties and refining and some of the things that you can’t do from an IT perspective.

But. I think for entrepreneurs, it’s an extraordinary opportunity. And those that will win. in my mind over the next few years are going to be the ones that leverage the tools most quickly because it’s not possible any longer to develop in the way that people were developing two or three years ago. If you do that, you’re going to be way late. And so now it’s about not so much your product, but learning what the state of the art is, which is literally changing every day in AI. And it’s a golden age, I think, for entrepreneurs. I think it’s going to be much, much more difficult for investors as an environment because the wealth of ideas are going to get much further along.

And that makes it more difficult, not less difficult, I think, to be an investor because you have more mature products. The entrepreneurs are going to be more mature and the entrepreneurs will have more leverage. And they may be able to make it to market much earlier than they would have otherwise, which means where they might have gone for a second round of seed capital, they may be able to get to market and be into revenue with a single small round of seed capital or no seed capital. And that makes that whole early, early stage ecosystem of angel and venture investing much more challenging. And so I think it’s just a great time. I do think that there’s a huge risk, and I don’t think it affects entrepreneurs or young entrepreneurs, but I do think there’s a huge risk of an overbuild.

It feels a lot like the leverage in terms of optimizing hardware and infrastructure is only going to get better, and it’s potentially going to leave us with actually a – I know right now we’re worried about power, we’re worried about compute, we’re worried about data centers, but I would project if we sat here two years from now, will be looking at a grand overbuild with a real challenge around ROI and how to make all these investments work. And so that’s going to be another positive for the entrepreneur because those resources are going to become very inexpensive relative to even what they are today. And so in that sense, I think it’s a great day for young entrepreneurs.

Ujjwal Kumar

Thanks, Jeff. Jeff, picking up from you about leaving from some of the AI tools, going to market faster, build out, ROIs, now we move to the right time, actually, when you spoke about ROIs. One of the things I was very curious about, all the world leaders coming here and putting a big bet on India. So, like, we just heard Sundar yesterday talking about $15 billion. New C cables between, like, with India. new innovation hubs. Rosalie, you are leading the clean energy transition with Google. I wanted to understand what Google’s, AI’s scale demand basically in terms of energy and why you are placing such heavy weight on India.

Vrushali Gaud

Okay, good. Thank you. Thank you all for joining and I appreciate it. I am being pitched against my boss, so I’m going to try and keep it as entertaining and as nice and valuable as I can. That’s a very interesting question in terms of the scale and why India. But I’ll build on a few things that both of you spoke about. One of the interesting things about this particular AI innovation timeframe you’re looking at is it’s what I call across the full stack. So you’re looking at a lot of things that are happening, which typically we talk about software, AI models, applications. That’s the shiny objects everybody talks about and very exciting. But then the amount of work that’s happening on underneath that, which is why I love this session, too, is beneath the AI.

The physical infrastructure layer of it is fascinating. And that goes from everything from the foundational layer that you’re talking about, which is your materials, your data center construction, your access to energy, to water, to all those foundations. So then how do you construct things the right way? We forget about the physical. These are all buildings, quite a few of them. How do you construct them the right way? And then how do you operate them the right way? And then the use of that. And so what we are seeing is just tremendous value and innovation across the entire stack of AI. Which I. Which I, as an engineer, find very, very fascinating. So in terms of Google, I think the.

The privilege and responsibility that Google has is how do we bring about the most value across that full stack, both from a business perspective but also from the impact perspective. And so a lot of the investments you’re seeing, you’re seeing across those pieces, right? So if you walked across this summit, you would hear different pieces of it. Our expo was mostly featured on the product side, so AI for education, AI for healthcare, AI for agriculture. Culture, how do you use AI in domains, contextualize it, and all of that has a layer of a country and where that context is. And then the announcements you talked about were a lot more on the physical side. So it’s what’s required for data centers.

You need good design, good builds, but then you also need network. And so the sub -C cable announcement is part of that. And if you read a little bit, it’s fascinating. It’s an India -America connection, but it goes one way we are building is across Africa. So that’s a big reason. It’s a big reason to bring on board. And then the other way it goes from Singapore and Australia. So it’s a fascinating network, which, again, you can only build data centers, but what’s the point if you can’t actually use it and network and bring them closer to wherever the edge cases are? So super excited about those pieces. Now going to your point about why India.

So why not India, I think, is what I would start with. But most people know it’s a billion -plus user. It’s a great growth market. It’s a lot of young population who we think are going to be the frontier of the growth. It’s a lot of population who also is very eager about tech and tech adaptation. So if you think about what happened with fintech and the phones and digital tech, a lot of the APAC countries, Asia and Global South, jumped ahead. I see people who didn’t even have credit cards. Now everybody uses GPay and UPI and all of those, right? So there’s a whole revolution where you can skip and build. And I think that’s another big exciting part of investing in India is can a generation of innovators come up?

Who don’t have the linear growth that we’ve seen in other regions but can leapfrog it? and I from an operational perspective feel super excited the same way about clean energy you can talk a little bit more Prince about that but India is one of the fewer places where the math on clean energy just works there’s growth, so there’s tremendous demand, lots of solar wind potential tremendous research going on in battery long duration storage, good policies and then the biggest issue what we’ve seen in the US is grid but they’re trying to build a high frequency grid which is fabulous, which then you bring in the innovation on that layer and that’s the unblock and if only you could solve permitting issues then you’re solving the whole stack that’s the excitement, it’s where the math works where the business case works, where you’ve got the talent and the innovation potential and then you also have the users

Ujjwal Kumar

wow, that’s amazing I do understand now thank you Thanks. So moving from that side, we heard about the demand side, and I’d love to take the insights from Prince, who is actually building the digital public infrastructure for the power sector, and they have been doing some incredible work, which I’ve seen in the past few weeks, particularly about P2P trading. And Prince, with all the initiatives about grid reforms and the trading platform which you are launching at this summit, how do you think the AI’s energy demand is going? How are you supporting it? Your insights.

Prince Dhawan

Thank you. Thank you, Joel. Thank you, everybody, for being here this morning. Let me first start by putting the AI. Thank you. compute demand in context. I honestly, resonating from what Duan had also said in his remarks, I feel that AI essentially will not scale unless your power is programmable. Okay, and that is, so the AI, I would say, I don’t want to call it race, but the AI build will depend a lot on, not on chips, as we might think. We do have the capacity and capability world over to solve that problem. But I think the binding constraint would be grids. It would be how intelligent and resilient your grids are. And I believe that is where, what is going to define the development of most of the compute infrastructure.

Now, what, India has essentially started doing is, we are redefining the architecture. Okay, so we are redefining how we view the grid. India already has one nation, one grid, which is essentially meaning one frequency. And now we are also having one digital interoperable layer that is being brought in by the India energy stack. So what does this mean? What does the stack essentially do? So the India energy stack, it basically creates the interoperable rails for systems to interact with each other. If you have a data center, it is not just creating high demand, but it is creating high peak demand that needs the grid to respond. And that is where you need coordination at scale.

So what is going to be scarce in the times to come is not electrification, as Roshani said. We have enough math works when you talk about solar power, when you talk about wind, even hydro. So that is where the math does work. But what needs to be ensured is coordination at scale. And that is what the India Energy Stack is essentially doing by laying down those foundational building blocks. Now, what we started with was a first showcase of how you can use the stack to essentially source energy from distributed energy resources like the solar rooftop panels, which we have on top of our households. We can literally transact in energy the same way that we transact using GPay, UPI payments.

Or using other such applications, Paytm, PhonePay, etc. So similarly, just imagine. that the data center, instead of relying on long -term PPAs and then hoping that the grid will deliver, can essentially source its power from millions of such distributed rooftop assets dynamically at scale. Just imagine the power of that happening. So it can literally be generating livelihoods for a lot of people who may not even be in geographical proximity to the data center. So individual retail households can essentially monetize their rooftop solar power by supplying to such data centers. How does the stack enable it? The stack lays down standard rules for measurement, identification, and settlement all in near real -time. So that’s how the architecture of the grid itself is changing.

let me the grid evolves generally in decades as to one said we have not invested heavily in the grids world over might be that China is an exception there but India has started doing the plumbing work it has started doing the hard work on that layer and generally AI evolves in quarters but the grid would evolve in decades how would you keep pace right and so that is where the India energy stack comes in where we push that development frontier and we enable people to talk to each other on the grid so AI would need not just electrons not just chips not just electrons it actually needs intelligent electrons and that is where the India energy stack sits in I think that should be in the times to come one of another reasons beyond economics that companies like Google or other companies would take bets on India.

And let me also tell you, you did recount Sundar’s message about $15 billion, but there’s also Reliance’s message about a trillion dollars in the next seven years. So let’s not forget that as well. I’m just putting stuff in context.

Ujjwal Kumar

Two minutes to one. li

Tuan Ho

ke India. And by the way, India is very, very well represented in Cambridge, which is how I probably met half the people on this panel. But it really does create these global scale opportunities to reinvent, to create this other, to support this other industrial revolution beyond just what the AI and the intelligence is allowing us to do. T

Ujjwal Kumar

hanks, Juan. Yeah, that’s exciting. Now with that, we want to move to Dr. Tobias. We have been talking about physical layer infrastructure. He has been working in semiconductor innovation since last 20 years, building it across US, Europe, India. I wanted to ask him, what does the next innovation looks like to you? Or what are you working on at this point? Where are you placing? your bet.

Tobias Helbig

Yeah, thank you very much for the question. Thank you so much for having me here. It’s great. And I would like to build a little bit, Jeff, on what you said earlier where you had this, are we on the right track? What the heck are we doing? And let’s zoom out for a moment. 1942, the head of IBM made a statement. There’s a world market for about five computers. And he was right, given the kind of computers he was looking at. We know better now, some years later, there’s laptops, PCs, there’s mobile phones, there’s basically a computer in every device. There’s billions of computers. Now what we discuss is, hey, AI, huge disruption. Power hungry like hell.

Shall we build some new computers? Shall we build power plants? Or how do we run it with renewable energy? and I get this nagging feeling is this really it or are we missing what came after these five computers in what we’re discussing if I take benchmarks like here’s my brain and it takes 20 watts there’s a fly which is a pretty agile intelligent robot below a milliwatt there’s something there’s something which is going to happen which is different than what we’re discussing here at the moment and that is what’s driving us as a semiconductor company in building on what starts now and driving it out into the real world so we today have products where on whatever 10 watts or so you can run very meaningful LLMs you can interact, you can drive the intelligent into the edge, into our real world that goes hand in hand with what’s happening around here moving from, hey, I can perceive something, is it a dog, a cat, to I can think, generative AI, I can create something out of those models to the point that I can create agents, stuff which acts on my behalf out there in the real field, which drives the intelligence and this disruption you’re looking at here at the moment and which is driving all these conversations, drives it close to us into the real world to the point that these devices, these robots, these whatever you want to call it, they’ll be able to learn.

So what we discuss here, and this is a huge challenge, I totally agree with all statements made before, we’ll see a next phase. It will see this moving into the real world, moving close to me, moving into autonomous systems, which ultimately change my life and change industries. And there is this second wave building up and my expectation, to some extent my hope, is from now on, where we sit as a company, is that this huge thing you’re already discussing with data centers is the five computers. And what is coming is these billions of edge devices which we will also see in the AI space. And just giving an example, I’m running marathons with a watch with me.

I charged it before I left in Germany and it still stays 12 days battery power. And there’s a lot of intelligence in that watch. This is where we are going. So the one is feed the beast and make it happen. The second is avoid that the beast is hungry and look at totally different models which will come in the next phase. Thank you.

Ujjwal Kumar

This is very interesting. Now we are talking about taking AI out of data center now. Any comments from the fellow panelists?

Jeff Binder

I agree with him. I think that the IBM analogies are very good. Very good one. I think we are all focused on the core and centralization. And as we’ve seen in many markets, they move from centralization to decentralization to hybrid approaches. And so that’s, I think, an incredibly astute observation. I do think edge devices ultimately have to be the core component in the full proliferation of AI. And so that means that, you know, as he said, small amounts of power can generate lots of value. It doesn’t necessarily have to be tokens in the center of a data center. So that goes to my concern that I think that – and look, I think all of the resources that are being built will eventually be consumed.

That’s not a – that’s a given. It’s a question of when and what – on what ROI they’ll deliver as they’re being – being consumed and used. And I think that’s a huge risk because – agents at the edge, which are probably going to end up being in the end a much more likely modality a decade from now. And it’ll be interesting to watch for sure.

Prince Dhawan

Okay. I do have a small – and I completely agree, actually. That’s truly, as Jeff said, it was an absolutely astute observation. But you know where you can see this being played out in practice even today? And that’s when you talk about finance. Okay. So finance world knows this. So today, because I work for a non -banking financial company, and one of our main products is infrastructure financing, where data centers are a product that we finance. And Roshali was in a panel discussion that spoke about the trifecta of AI energy and finance. But you know the finance bros, they have figured it out because today if you go for financing of a data center, you won’t get debt financing for GPUs.

GPUs are mostly financed by equity because there is obsolescence risk in GPUs. You would get debt financing for the brake motor, maybe even for sourcing power, but you won’t get debt financing for GPUs. And there you have it because they are seeing the big picture being played out there. So completely on board there, yes.

Ujjwal Kumar

Again, Rukhsani has been…

Vrushali Gaud

No, no, no, I’m good.

Ujjwal Kumar

No, no, no. We want to hear from you. Please, go ahead.

Vrushali Gaud

No, I think the risk of strata assets, the way you said, and the ROI is real, like in a sense of where you’re investing and what. But I think your point is very astute. There are portions of this will be obsolete. There are portions of this which will be very easily replaced, whether it’s on the chip side or whether how you write the programs. Even the models, right? You went from large scale, smaller. How do you build them? but also what I’m hoping is the bets on some of the hard infrastructure are just good things to do like I think to me the fact that we are seeing a transition to renewables or seeing a transition of the grids being operated in a better way, some of the boring bits that people didn’t pay attention to is how do you run things efficiently those I think are good pieces of this and then it goes to you right size it, we’ll get over the FOMO and the extra investments and it’ll probably get right sized into where in the stack you really want to invest with the ROI

Ujjwal Kumar

Thank you, so with this I’d like to take it a little bit more deeper, like we spoke about some of the opportunities we agree on something we may not on some Tuan, you invest early on founders I wanted to check with you and understand with you how do you, can you tell us that is there a mismatch between what is getting funded and what needs to be funded?

Tuan Ho

That’s a good question. Is there a mismatch between what’s getting funded and what needs to be funded? Well, I mean, probably. Yeah, I mean, I think, well, okay, going to the theme of this, I think there is more likely to be a mismatch between what is getting funded in the sort of like the pure AI world, if we’re talking about the foundational models. I think, Jeff, I think you had made this comment a little bit earlier. You look at a lot of the AI companies out there, and it’s a little bit like the dot -com era where you’ll see 100 companies, and the reality is that in five years, there will be five of them that are left.

I think one reason why I like focusing on infrastructure -type businesses is because I think… I think there’s more durability. and clarity to exactly what the problems are that you’re trying to solve. I mean, every great startup begins with a really well -understood problem and a product, what they call a product -market fit, like a founder that’s able to build a great solution to that problem that has some sort of market validation in need. And what I find really exciting about infrastructure businesses is I think the problems are a lot clearer in terms of what you’re trying to solve. To your point, there’s a lot more risk in the GPUs. There’s a lot more risk in the models that you’re building, that you’re building around them.

And the reality, too, is that those things are also going to change a lot faster. I mean, if you look at a data center, as an example, a data center ultimately is a giant box that provides a lot of power at scale and it needs to be able to efficiently… efficiently cool what’s inside it. of it. In terms of what GPUs or compute you put inside, I mean, that can change over many, many generations. But the utility of the infrastructure you’ve built there will always have value. So, I don’t know if that answers your question.

Jeff Binder

To add to that, I think that if you look at the dot -com era, measuring, with the exception of hardware companies, which were in switches like Cisco and other players, it was very difficult to determine whether a product was good or not. For those who remember MySpace before Facebook, it looked like MySpace was going to own the social media space. Of course, most, half the people in here probably don’t even know who MySpace is. It’s much different now. There’s a measurability component in all aspects of AI that didn’t exist in the dot -com era. You know, you had commerce platforms, but it wasn’t clear what made one commerce platform better than another. The consumer would ultimately decide that over time and through iterations.

And if you remember, Amazon for a long time was known for one -click ordering. Well, none of us really want to do that because we don’t want to make a mistake and find out that we bought the wrong thing. I think now it’s different. Almost every aspect of artificial intelligence deployment from the foundational aspects all the way to the top of the stack are measurable. And so that’s going to make the success and failure of businesses much more clear, much sooner than it was in the case of the Doctomer. And I think that’s going to be ultimately the element that shakes out companies very quickly. And then to the point about obsolescence and GPUs, we don’t know what the hardware roadmaps look like, even inside of a Google or a Jensen’s company, NVIDIA.

Or somebody else that’s out there. And power, which is the fundamental thing I think we’re talking about foundationally. can be grossly disrupted by those advances because if somebody has a breakthrough on chip design that’s now 10 or 50 or 100x what somebody else deployed, their data center is now almost instantly, at least from a financing perspective, obsolete. And so that’s a huge danger, I think, for investors in those foundational areas.

Ujjwal Kumar

Thank you. Dr. Tobias, you have also been involved in the innovation ecosystem very strongly. What is your take on this? What are you seeing because you are also involved in India? I’ve seen your company running hackathons and competitions. I’d love to know more from you.

Tobias Helbig

Adding to what just was discussed, we have a tendency to overestimate the next two years and impact and underestimate what’s happening in 10 years. And at the moment, we are going into this with huge bang. which even maybe have these ups and downs things even much bigger. From my perspective, all what we are discussing on AI is absolutely real. This is a huge disruption. This is changing industries. This is changing lives. This is changing professions. Wherever there is data, there is change. And that in the end is driving what we are doing by developing the products we have, which is semiconductors products, by being in India for that since literally decades. Development centers on our DNA history as a company of Motorola, Freescale, NXP, here in the Noida, Delhi region, in Bangalore, and so on.

So very much working on that. And on your question from an innovation perspective, well, we all know the hype cycle. And that’s tough. Because it always means that there is disruption. And there is a trap of disillusionment. And we’ve seen it. for all major breakthroughs, especially when they are being hyped up like hell. There’s the self -driving cars. There’s other things. In the end, these things get real. They have the substance. They happen, and they transform things. And AI will. The way there, and also in the question, hey, what’s the risk? What’s the bad, and what’s coming from the sidelines? I think we will see still troughs of disillusionments and surprises. There was one some while ago when this wave had a deep -seating moment.

Such moments will come again. And there will be a recovery from that, I’m also sure. So I’m in innovation since literally decades. I love it. It’s a roller coaster. We overestimate, we get shocked, and we get it right.

Ujjwal Kumar

Thank you. With that, I’ll go to Sari. I was very excited when I saw Google Logo launching Google Climate Technology Center. Okay. And I would like you to quickly give your insights, like what is it about and what would the innovators be looking for it?

Vrushali Gaud

Yes, thank you. So super excited. This week we announced in partnership with the Office of Principal Scientific Advisory for the Government of India, Google’s Center for Climate Tech. So there’s a couple, you know, how did we get here? Because that’s interesting to you is we see a lot of innovation. I live in Silicon Valley. I was raised in India, across back and forth. I’ve lived across the world. There’s different innovation which comes from big institutes, big academic settings, big companies. But there’s also innovation that comes from different corners of the world. What we loved about the PSA philosophy was they’re trying to get more Tier 2 and Tier 3 cities and also a wider spread of universities and academia that can get involved in this.

So, you know, besides your premier one. So that was very enticing. The other thing is, how do we take innovation down to the root? which I think also helps with some of the hype cycle because you’re making it local and you’re also making contextual to where those cities are. So with that in mind, what our center is looking for, we have a couple big pillars. One is skilling. We think there’s green skilling. A lot of focus on AI skilling, but in terms of green skills, which are decarbonization, clean energy, in terms of just we are looking at materials, chemistry, there’s a lot of new things in those spaces which haven’t been brought into college curriculum or university curriculum.

So we want to build upon that. A lot of the construction and investments are happening in tier two cities, so we think it’s a great way to get a more diverse pool skill in that. So that’s number one pillar. The second one is low carbon materials. So you go to embodied carbon, something you’re all very passionate about. So how do you drive innovation in construction, which is going to be huge? And again, it’s not just data centers. What you learn from data centers can be for real estate, for commercial buildings. So it’s to do with low -carbon steel, low -carbon cement, and low -carbon materials as you see them go through that construction cycle. And the third one we have looked at is right now that we have is sustainable aviation fuel, which is a little different from data centers.

It’s not that, but I think it’s like a good growing area, which we are, again, one of the philosophies we have is where can we find first -of -a -kind pilots and places where we can build, bring the Google brand and innovation. And we think sustainable aviation fuel in a growing country that is like now has one of the fastest growing airports and air traffic, that would be a good one too. And our hope is as we go through this, we are trying to see very outcomes -based, so not pure research, but pilots and actual update.

Ujjwal Kumar

Thank you. Very quickly, Tuan, do you have any closing 30 seconds?

Tuan Ho

I was going to say there, one thing that I don’t think we had a chance to do. We didn’t discuss as much, but it is important. especially as we’re at an event like this is government financing. I think what’s really another thing that’s been really exciting about this is having a tech conference like this where you have the Prime Minister and multiple heads of states coming from around the world to say like, these are things that we need to invest in, these are things that we need to support, is I think from a tech VC side of things something that we’re not used to seeing. But I also think it’s very exciting. Both in the United States you’re seeing hundreds of billions, hundreds of billions of dollars being invested by the federal government into infrastructure.

And you’re seeing similar investments being made in countries like India, countries outside of, China’s been doing this for a while, but you’re seeing this happen around the world. And so yeah, I think, I mean, where are we right now? There is the . industrial, like I said at the beginning, there’s the industrial revolution that AI is ushering in, but there’s also the industrial revolution that the requirements of AI are also going to require or is also going to usher in. So I think it’s going to be a bright future for us all.

Ujjwal Kumar

Thank you. I think that’s a great closing for us, and I enjoyed talking to all of you. I really had so much fun. Thanks, and your insights are amazing. Hopefully the innovators looking here, they got something out of it, and we’ll see some new people coming to all of us doing the innovations. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you.

Ujjwal Kumar

Speech speed

64 words per minute

Speech length

AI Infrastructure & Critical Minerals Corridor

Topics

Environmental impacts | Financial mechanisms | Artificial intelligence

Tuan Ho

Speech speed

149 words per minute

Speech length

AI Infrastructure & Critical Minerals Corridor

Topics

Environmental impacts | Financial mechanisms | The enabling environment for digital development

Jeff Binder

Speech speed

148 words per minute

Speech length

Investment Risks, ROI & Funding Mismatch

Topics

Artificial intelligence | Financial mechanisms | Building confidence and security in the use of ICTs

Vrushali Gaud

Speech speed

189 words per minute

Speech length

Grid Modernization & Energy Stack for AI Power

Topics

Environmental impacts | The enabling environment for digital development

Prince Dhawan

Speech speed

129 words per minute

Speech length

Grid Modernization & Energy Stack for AI Power

Topics

Environmental impacts | The enabling environment for digital development

Tobias Helbig

Speech speed

150 words per minute

Speech length

874 words

Speech time

348 seconds

Second AI wave: billions of low‑power edge devices

Explanation

Tobias predicts that after the current data‑center‑centric era, AI will move to a massive fleet of edge devices that can run meaningful models locally.

Evidence

“And what is coming is these billions of edge devices which we will also see in the AI space” [79]. “There are billions of computers” [83].

Major discussion point

Semiconductor & Edge AI Future

Topics

Artificial intelligence | The enabling environment for digital development

From five computers to real‑world agents

Explanation

He describes the evolution from early AI hardware (“five computers”) to low‑power chips that enable autonomous agents operating in the physical world.

Evidence

“…this huge thing you’re already discussing with data centers is the five computers” [80]. “…on whatever 10 watts or so you can run very meaningful LLMs… drive the intelligence into the edge, into our real world” [80].

Major discussion point

Semiconductor & Edge AI Future

Topics

Artificial intelligence | Environmental impacts

Participant

Speech speed

56 words per minute

Speech length

156 words

Speech time

166 seconds

Foundational layer (critical minerals, energy, semiconductors) is key for AI

Explanation

The participant notes that the foundational problems beneath AI—critical minerals, energy, and semiconductors—are essential to address for scaling AI.

Evidence

“Quantum Alliance works with universities, industry and governments to get top talent working on the foundational problems beneath AI, from critical minerals to energy to semiconductors” [4]. “This is where the real opportunity lives” [10].

Major discussion point

AI Infrastructure & Critical Minerals Corridor

Topics

Environmental impacts | Artificial intelligence | The enabling environment for digital development

Agreements

Agreement points

Infrastructure businesses offer more durability and clearer problems than AI models

Speakers

– Tuan Ho
– Jeff Binder
– Prince Dhawan

Arguments

Infrastructure businesses offer more durability and clearer problem definition than pure AI models

Market evolution follows centralization to decentralization to hybrid approaches pattern

Finance sector already recognizes obsolescence risk by refusing debt financing for GPUs

Summary

All three speakers agree that physical infrastructure investments are more durable and less risky than AI models/GPUs, with the finance sector already recognizing this by providing debt financing for infrastructure but not for GPUs due to obsolescence risk

Topics

Financial mechanisms | The enabling environment for digital development | Artificial intelligence

AI will evolve from centralized data centers to distributed edge computing

Speakers

– Jeff Binder
– Tobias Helbig

Arguments

Arguments

Measurability in AI makes success/failure clearer and faster than dot-com era

Hype cycles create troughs of disillusionment but real transformation ultimately occurs

Summary

Both speakers acknowledge that while AI follows typical hype cycles with periods of disillusionment, the measurable nature of AI deployment makes business success and failure clearer than in previous technology booms like the dot-com era

Topics

Artificial intelligence | The enabling environment for digital development | Monitoring and measurement

The enabling environment for digital development | Financial mechanisms | Artificial intelligence

Unexpected consensus

– Prince Dhawan
– Jeff Binder

Arguments

Finance sector already recognizes obsolescence risk by refusing debt financing for GPUs

Risk of infrastructure overbuild similar to fiber buildout, but resources will eventually be consumed

Explanation

Prince uses finance sector behavior to validate infrastructure durability, while Jeff warns of potential obsolescence from hardware breakthroughs that could make data centers instantly obsolete – an unexpected contradiction in risk assessment

Topics

Financial mechanisms | Artificial intelligence | The enabling environment for digital development

Innovation timeline expectations

Speakers

– Tobias Helbig
– Ujjwal Kumar

Arguments

Hype cycles create troughs of disillusionment but real transformation ultimately occurs

AI is forcing creative destruction of global infrastructure across energy, semiconductors, and critical minerals

Explanation

Tobias emphasizes caution about hype cycles and overestimation, while Ujjwal presents AI transformation as currently happening at unprecedented scale – unexpected disagreement on transformation timing

Topics

Artificial intelligence | The enabling environment for digital development | Innovation and Technology Development

Overall assessment

Summary

Main disagreements center on infrastructure investment timing and risk, computing architecture evolution, and primary constraints for AI scaling

Disagreement level

Topics

Financial mechanisms | Artificial intelligence | The enabling environment for digital development

The enabling environment for digital development | Financial mechanisms | Artificial intelligence

Takeaways

Key takeaways

AI scaling requires a complete infrastructure revolution spanning critical minerals, energy systems, semiconductors, and grid modernization – representing the largest infrastructure buildout in human history

The binding constraint for AI development will be intelligent, programmable power grids rather than chips or compute capacity

Current centralized data center approach represents an early phase, with the future moving toward billions of edge AI devices requiring minimal power consumption

India emerges as a strategic AI hub due to its billion-plus user base, leapfrog innovation potential, favorable clean energy economics, and digital public infrastructure capabilities

Infrastructure investments offer more durability and clearer problem definitions compared to pure AI model companies, which face higher obsolescence risks

AI tools are democratizing entrepreneurship by reducing capital requirements and enabling faster go-to-market strategies with unprecedented leverage

Government financing at unprecedented scale (hundreds of billions globally) is supporting this infrastructure transformation

Innovation must occur across the full technology stack from foundational materials to applications, not just software layers

Risk of infrastructure overbuild exists similar to the dot-com fiber buildout, but resources will eventually be consumed as the market matures

Resolutions and action items

Google announced $15 billion investment in India including gigawatt-scale AI hub in Vizag and four new subsea cables between US and India

Launch of Google Climate Technology Center in partnership with India’s Office of Principal Scientific Advisory focusing on green skilling, low-carbon materials, and sustainable aviation fuel

India Energy Stack implementation enabling peer-to-peer energy trading for data centers to source power from distributed rooftop solar assets

54 countries launched FORGE framework for AI-powering minerals coordination

US-India critical minerals corridor development to reduce dependency on China for rare earth magnets

Unresolved issues

Mismatch between current funding patterns (favoring AI models) versus actual infrastructure needs

Uncertainty about hardware roadmaps and potential breakthrough technologies that could make current investments obsolete

Timeline disconnect between AI evolution (quarters) and grid infrastructure development (decades)

ROI challenges for massive infrastructure investments given rapid technological change

Permitting and regulatory issues that could block clean energy infrastructure deployment

How to balance centralized data center investments with emerging edge computing requirements

Financing models for high-obsolescence components like GPUs versus durable infrastructure

Suggested compromises

Hybrid approach combining centralized and decentralized AI computing to balance current needs with future edge requirements

Focus infrastructure investments on durable components (power, cooling, buildings) while treating compute components as equity-financed due to obsolescence risk

Leverage cross-border talent collaboration using AI tools to bridge cultural and technical gaps between US and India

This insight distinguishes the current AI boom from the dot-com era by highlighting the measurability of AI performance, suggesting that unlike the subjective nature of early internet products, AI solutions can be objectively evaluated, leading to faster market validation or rejection.

Impact

This comment added analytical depth to the discussion about investment risks and market dynamics. It provided a framework for understanding why the current AI infrastructure buildout might be different from previous technology bubbles, influencing how other panelists discussed the sustainability and evolution of AI investments.

Overall assessment

Explanation

The observation that GPUs get equity financing while basic infrastructure gets debt financing suggests a need for more sophisticated financing approaches

Disclaimer: This is not an official session record. DiploAI generates these resources from audiovisual recordings, and they are presented as-is, including potential errors. Due to logistical challenges, such as discrepancies in audio/video or transcripts, names may be misspelled. We strive for accuracy to the best of our ability.