Who Holds the Risk in AI Infrastructure?

Tonya Bakritzes: Right now, a huge amount of capital is flowing into AI infrastructure. Hyperscalers alone are on track to spend $750 billion this year on chips, data centers, and the power behind them.

Spending on that scale has a lot of people worried about a bubble. Karen Fang sees it differently. With power, equipment, labor, and permits all in short supply, she argues we’re likely underbuilding.

But someone still has to pay for all of this. More and more of it is financed with debt and outside investors, which makes AI a clear window into one of the forces we track at S2G: the price of money. There’s plenty of capital, but not every project gets it on the same terms. And data centers are especially hard to price. They are tied to one grid, one water supply, and a set of local climate risks for decades.

In this episode, Sanjeev Krishnan sits down with Karen Fang, who is the Global Head of Infrastructure & Sustainable Finance at Bank of America and co-heads the bank’s global capital solutions business. They talk about how higher borrowing costs are showing up in data center deals, who’s actually holding the risk on the buildings and the chips inside them, and why most new electricity on the U.S. grid is still coming from solar and batteries.

For climate investors, the tailwind has shifted from the IRA to electricity demand. Can we finance this buildout and everything else we need at the same time?

Please stay tuned for important disclosures at the end of this episode.

Sanjeev Krishnan: Karen, welcome to the S2G Podcast. We talk a lot, and I’m excited that our listeners can hear from you and experience the dialogue I get to have with you every once in a while.

Karen Fang: That’s very kind. Thank you for having me, Sanjeev.

Sanjeev Krishnan: You’re at the center of so many interesting things, and I want to get to all of them. But I always love the part in When Harry Met Sally when they’re driving from the University of Chicago to New York, and I think it’s Harry that asked Sally, or Sally that asked Harry, tell me the story of your life. So tell me, where did you grow up? How did you end up in finance?

Karen Fang: Sure. I was born and raised in China until I was seventeen years old. And I went to Tokyo because I couldn’t come to the US. My parents, who were professors, told me that the US is simply too far from China. I studied Japanese for two months and got into the University of Tokyo. I studied economics and financial engineering. I got a bunch of offers in my junior year at the University of Tokyo, which is obviously one of the really good universities in Asia, all from foreign banks. I didn’t apply to any Japanese companies and I ended up at Merrill Lynch, Tokyo. And my first job was equity derivatives trading, one of the very few young and female traders on the desk. 

And 1998 comes around, that was my first year, the LTCM crisis happens. And I somehow got into derivatives, correlation derivatives, quanto hedging. Don’t worry if you don’t know what those wonky terms are. Long story short, I was fairly successful at it and realized I really like structuring instead of trading. I moved to Merrill Lynch in London and then came to the US in 2002. So I’m definitely a New Yorker, living in New York City for 24 years, always in finance. And went to Goldman and came back to Bank of America Merrill Lynch in 2010. So 16 years on this tour of duty, and always in structured finance, complex solutions. If you don’t have a solution, I’m the woman that you call.

So it’s been pretty interesting, and my journey kind of landed me in sustainable infrastructure, which I’m still passionate about today, and project finance, infrastructure finance. Given the digital infrastructure revolution that we’re experiencing right now, Bank of America, expanded my team and my mandate beyond sustainable infrastructure officially as of two years ago to all energy and power infrastructure, all digital infrastructure, and core infrastructure. I’m also co-heading our global capital solutions business, which has become increasingly important because capital demand in terms of speed and scale is just so unprecedented.

Sanjeev Krishnan: I know some of the long-term capital stuff started with some of the Asian currencies.

Karen Fang: Yes.

Sanjeev Krishnan: So were you on the front lines of some of that?

Karen Fang: Yes. I joined on April 1, 1998. So when LTCM and Asia, you know, the currency crisis and all of that happened, I was on the trading floor of Merrill Lynch, Tokyo, which obviously, suffice to say, I learned what volatility was and what disruption was and how you actually need to hedge a lot of different risk factors very quickly. And tail risk is not a tail risk if you actually think carefully about it.

Sanjeev Krishnan: I was trained in financial services my whole life. And one of the things that I don’t know if it’s accurate, but the way that I’ve always mentally modeled this is a lot of trading and markets are relative value flows for lack of a better word and a lot of corporate finance and capital solutions, and what you do is fundamental value. A, do you agree with that mental model? And then B, the fact that you did both, does that help you in navigating the flow side of it versus the stock? How valuable was it to come up on the market side? 

Karen Fang: I think coming up from the market side and the sales and trading side just means you actually have a shorter time frame to think about everything. And the relative value aspect, frankly, applies to all. When you’re thinking about a data center financing with a hyperscaler as a tenant versus a corporate bond linked to that hyperscaler, there’s relative value inherently in that play.

Growing up in that fast-paced environment, where you’re always thinking about two- or three-standard-deviation moves and hedging tail risk, is really good training. Now, when we are thinking about financing a 15- to 20-year project, we have to consider what could go wrong, but I just frankly have a little bit more time to think about it.

Sanjeev Krishnan: Anyone who entered any sort of financial services in the late ​’90s, I think, grew up with a series of things that we were told never happened before, right? You know, even forget sort of the Asian contagion for a second, but 9 – 11, ​’07 – 08, really ​’08 – 09.

Karen Fang: The TMT crisis in 2000. 

Sanjeev Krishnan: Yeah. The tech bubble and then COVID, then Russia-Ukraine, and what’s happening now geopolitically. What I think about a lot is that we’re building S2G around these five megatrends or transitions — CPU to GPU, unipolarity to multipolarity, geopolitics, a declining price of money to an unknown price of money, 1.5 to 2.5 degrees of warming, and from a demographic dividend to aging. We’re trying to figure out how you underwrite all five and create outperformance from an asset class and investment perspective. At the scale that you operate in and at the breadth, do you see those five as relevant in your world? Are there other megatrends you’re seeing that you think will impact capital solutions or infrastructure? How do you think about the megatrends?

Karen Fang: For sure. I completely agree with the five you outlined. I think the other megatrend, which we also have to be cognizant of, and again, the good news is that we have time to think about it, is whether we are looking at a sustained inflation cycle, given supply chain disruption, geopolitics, and, inherently, just frankly, certain aspects of being very supply-constrained. Again, infrastructure space, my space is seeing that every day. Sustained inflation, that’s a megatrend and actually gets into the government deficit across the world. And that is the base rate for all financings, and that is a risk that we, perhaps as a market, mispriced a bit a year ago. 

Those are fundamental megatrends as well, including government deficits, asset-liability management, and public-budget management, and what’s going to happen to Social Security.

Sanjeev Krishnan: All the things we kicked the can down the road because it was politically easy and central bankers felt like it was too big to fail or didn’t want to absorb the pain. I think in the next thirty years that’s going to come home to roost in some transmission mechanism. 

Karen Fang: Maybe sooner.

Sanjeev Krishnan: Maybe sooner. Yeah. I didn’t want to be that alarmist, but I agree with you. So the view is that we’re moving from an era of moderation to one of volatility. I do think the point you just brought up is super interesting. Over the last 30 years, inflation has been a flow dynamic where I think you could set rates on the base rate or the Fed window, and then, effectively, you had, you know, the Pearl River Delta that kept consumer prices in check, and a demographic dynamic that kept labor in check. And geopolitically, there wasn’t this level of geopolitical volatility. You have sixty-one wars, the most state-sponsored wars since World War II. And so you had a sort of benign second-order dynamics that I would call flow inflation. 

What I worry we have is feedstock-driven inflation, with the biggest CapEx cycle probably since the railroads, and you’re in the middle of a lot of it. And nothing that the FOMC, the Fed, or the ECB does can produce more HVAC contractors or plumbers. Nothing they can do to address rare-earth mineral processing or other critical-commodity dynamics. They really can’t solve diesel prices or some of the tariff dynamics that are occurring. And so it’s an interesting time to talk about inflation and the role that monetary authorities have in tackling their nail and hammer. And you seem to be alluding to that.

Karen Fang: Absolutely. And I think what we are perhaps starting to realize as a market and as an ecosystem is that this AI economy is very different from the internet era. People like to compare the two, right? One is software-driven; the other is hard hardware or physical-aspect-driven. The AI economy and the AI constraint that we’re talking about in terms of compute is constrained by physical access to power, cooling, equipment, and labor, as you pointed out. The AI renaissance is so physically constrained by everything, including interconnection queues, including the community opposition we’re seeing right now, regulatory approvals, and all the permitting.

The technology could grow a lot faster if it was like the internet era, where it’s a software-driven revolution, and it’s high margin. CapEx is decent, but it’s not crazy. And here we’re looking, for this phase at least, lower margin, because in order to produce the compute power, you need the power, you need to build it from scratch. And the margin is lower, and the risk is higher. And then you’re massively under-supplying right now, given the physical constraints that we just talked about. So I think the two phases, comparing ourselves to the 2000 TMT era, the internet era, are just very wrong in the sense that we were not constrained much by physical world constraints at the time.

Sanjeev Krishnan: I’ve been trying to study general-purpose technologies over the last three hundred years. When it [GPT] went from invention to productivity, that lag, it’s sometimes ten to thirty years, right? Between the steam engine and the productivity hit on the economy, from the ICE engine to the productivity hit. And often I think about, I started my career in TMT M&A, and we built a ton of fiber and then didn’t really need to build it again for nine or ten years.

Karen Fang: Given the amount of constraints we have and given that we’re probably realistically speaking building about fifteen, twenty gigawatts of data centers in this country right now, compared to hundreds of gigawatts of data center announcements that have been made, one can reasonably argue that we are underbuilding it right now. Therefore, the oversupply risk, and therefore, the obsolescence risk of a data center that’s going to become a stranded asset, I think right now, is relatively low. Meanwhile, we have time for the technology to catch up. Chip efficiency is going to improve, compute efficiency is going to improve. This is my favorite topic: what AI can do in matching energy supply and demand, making sure we’re not wasting power that we produce, and adding a lot more batteries, more efficient battery technology to unlock that excess power. There is actually time for technology to catch up and hopefully reduce the energy and power requirement to produce the same unit of compute as we deal with the constraints and deal with all the permitting and delays.

Sanjeev Krishnan: I want to talk about the price of money because, for a long time, when rates were zero or even negative, infrastructure became the modern bond. Effectively, people got much of their yield from buying infrastructure-related products as opposed to Treasuries or other similar products. Maybe just unpack what’s happening in the Treasury market and perhaps its impact on infrastructure markets.

Karen Fang: Sure. I think the fact that we are seeing a sustained risk premium priced into the Treasury rates right now is a concern about sustained inflation, which we talked about, as well as perhaps the lack of confidence that the budget deficit and a sustained positive surplus in our budget is achievable.

So when you have that, you’re foreseeing a tremendous amount of supply coming in because we simply have too much of a current account shortfall, and the gap has to be filled with additional Treasury issuance. I think what we’re seeing right now isn’t just the shape of the Treasury curve; it’s that the overall risk premium is elevated, and we don’t seem able to address the geopolitical issue. We don’t see oil and energy prices coming down anytime soon, even with the Venezuela news. Ukraine-Russia is still happening, obviously, you know, the Iranian situation is still happening. So, clearly, we don’t see the energy-elevated price issue being resolved anytime soon. But it’s not just energy, it’s really the supply chain. Frankly, the cost of goods and services overall is pretty elevated, given that there is a basic commodity input into everything we consume.

I do think that the market has now resigned itself to the fact that this is not going to be a short-term resolution; it’s going to be a long-term issue, and it may take time to resolve. And meanwhile, you know, we have a lot of spending. We don’t have a revenue problem. As some would say, we have a spending problem. So we don’t know, right? As a capital market, as an investor, you want to be compensated for that risk premium.

I don’t see that problem resolving itself anytime soon. And a lot of the CFOs and treasurers of large companies globally are now saying, okay, how do we think about the base rate at 5% and adding our credit spreads to it in terms of my cost of funding on the debt side and in terms of equity. Equity tends to be priced at a risk premium to risk-free Treasury rates, so the overall cost of capital is just going to be more expensive for longer. So can you catch up then, right? If this is on the expense side in terms of cost of capital and cost of goods, can you actually make more? 

And we’re seeing that. A direct example of that expression is: if a data center tenant was going to lease a data center for 15 years or 20 years, we saw a year ago a lease-to-cost in terms of lease yield at X. Now we’re probably seeing X plus 100 or 150 basis points priced in. So, if you are pricing an infrastructure project and you’re asking, ​“What is my revenue?” The lease yield has increased, and so has your cost of capital, so that is obviously elevating at the same time. And projects that didn’t have a lease renegotiation mechanism, an inflation escalator, or some form of price resolution are seeing issues with revenue versus expenses and potentially with your financing mechanism.

The price of money is expressed every single minute in the capital markets. Not every project is made equal, not every company is made equal, not every government is made equal. But the overall sentiment is that we’ll be in this environment for longer. And the overall cost of capital will be higher. And therefore, what are some of the solutions that we can provide, whether it’s on the revenue side, whether it’s on the diversification side, or, you know, frankly, a selection of your inputs on your CapEx to where you can save on the insurance and everything else.

Sanjeev Krishnan: Where we’re trying to play is, can we bring deflation into a market that’s highly inflationary? So can our companies offer goods and services that are deflationary in a highly inflationary environment? And if you can produce a good ROE and good cash flow from offering that deflation, you can obviously win in the marketplace, which is one way that I’m thinking about it. Obviously, it requires a combination of innovation and business model innovation to offer deflation, given all the other sorts of complexities we’ve outlined. But when you can find a company that can do that, we want to be super aggressive.

Karen Fang: Well, one way to kind of produce this deflation and potentially solve some of the input problems on CapEx, right, is where you see the most amount of constraints. Gas turbines, switchgear, transformers, and cooling equipment, all of these are constraints. So sometimes I wonder, again, this is not like a perfectly thought-out solution, how do you produce that deflation? You have to increase supply. So how do you increase the supply of all of those extremely constrained parts. 

It’s potentially industrial policy. When I think about COVID, what happened ultimately, whether it’s ventilators and PPE equipment, masks, as well as ultimately vaccines, we had an industrial policy to make sure all of our focus and resources are pointed in one direction to solve a crisis. Perhaps we’re not talking about this as a crisis, and it’s actually being talked about as an investment. That’s not attracting the industrial policy to produce that oversupply, to drive deflation, to ease the bottleneck. 

But maybe there is something there to think about in terms of whether this is a moment of inflection. That we need this to stay competitive, to create jobs, to drive economic growth, because so much of the economy and so many consumers’ lives are going to be transformed by this technology, which is universal in utility. So it’s not just about a TMT, and the internet is nice to have. We actually cannot go back. 

Sanjeev Krishnan: Yeah. One of the interesting things that I think you’re finding, whether directly or indirectly, every day, is the social license side of this. I think social license, paired with the grand strategy, seems like what the moment calls for. I actually think what we did in COVID was both. We did social license and grand strategy. Now, some of the tail of how we did it, I think, still lingers. And no one’s going to do this perfectly. But if you don’t do it, you’re sleepwalking into a set of outcomes that are inflationary, uncoordinated, and perhaps offer both real economy volatility and financial volatility, is what I’m hearing. 

Karen Fang: Absolutely. I think we do need guardrails. I mean, the frontier labs are aligned on this. 

We should be thinking about safeguards to make sure we don’t lose control of AI without a human in the loop. It’s clearly not an ideal outcome right now.

Sanjeev Krishnan: The one thing I wanted to ask you, just with your current hat, is: people talk about data centers as an asset class now. I think about aviation finance and engine financing. That to me, you’ve got a couple of Rolls-Royce, GE, you have a few engine companies that can now finance this stuff with investment-grade, get it rated. Is compute becoming an asset class, and does it look like airline engine financing, or does it look different. 

Karen Fang: It’s a great question, and I think it’s easy and perhaps a lazy thing to do just to say, it’s a new asset class. But if you look at what is happening in capital markets right now and how these things are getting financed, and by the way, we’ve never done this much volume in my 28-year career, I would just stress that this is fundamentally different than anything else we have ever done in capital markets. 

Two years ago, if you did an infrastructure project finance, you did a two-billion-dollar deal; it’s a big deal. Now $20 billion is a medium-sized deal. I just want to stress there are profound implications of what this means in terms of asset allocation, in terms of whether you call this an asset class or you call this a relative value play.

The reason I say that is that it’s not straightforward. The answer is not straightforward. The equity investor in a data center or in a GPU-as-a-service project, perhaps, is viewing this as an asset class. The debt investor, which could be anywhere between 70 to 90 percent of the capital structure on any given project, is viewing this largely as a relative play to the underlying tenant or the lessee. 

So when you think about that, the reason I say that is, this is actually where, fundamentally, I think some of the mispricing of money can happen. If everyone says it’s a hyperscaler ABC, I’ll add a hundred basis points as a debt investor, and that would be my golden pricing rule of thumb.

Not every project is made equal, not every hyperscaler is made equal, not every developer is equal, not every asset location is equal, and not every insurance strategy is equal. So this is where we in Global Capital Solutions spend a lot of time thinking about what is the underlying asset when it’s the cash flow that you’re financing.

The reason I say the equity investor has to take a different view is that the debt investors, largely right now for both chips and data center financing, are looking at an amortizing sinking fund structure, an amortizing bond, where the debt principal is amortized to zero in the first lease term. 

We’re not taking any residual value risk of the data center in 15 years. We’re not taking any residual value risk of the chips in five years. That’s sort of the standard term. Where the residual value risk is passed on through the equity investor. So the equity investor has to take a view. It just happens to be that there’s so much appetite for this right now. The equity investor’s IRR is sufficient. So they can take that view: even if the residual value of the data center is zero in 15 years, they’ll still earn sufficient returns above their principal return.

That’s what we’re seeing right now as the pricing becomes more expensive. We talked about the base rates, and we talked about the risk premium. At some point, the equity investor is going to start taking more and more risk. And they have to take a more informed view of the residual value of that data center in that location, and perhaps be subject to physical climate risk in certain states or locations.

And the same thing on chips. The reason that I think Nvidia made that announcement is that right now, a lot of the chip financing, at least for the senior debt part of the capital structure, is relying on residual value support from the chipmakers. If that goes away, because that can’t be forever, then you have to take a view on the value of the chips in five years, seven years. And that is required to happen because, as you know, right, every year we’re talking about two to three trillion dollars of CapEx, and then a big chunk of that is chips.

You can’t have investment-grade chipmakers guaranteeing the residual value of chips forever. It’s trillions and trillions of dollars. Yes, they’re all trillion-dollar companies in terms of market cap, but there is a limit on how much they could do. To bring it back to your last point, aircraft, just like autos, have a very proven observable residual-value market. An airplane that’s 20 years old versus 30 years old, you know where it’s going to trade in the market; there’s liquidity for that.

An auto after a three-year lease term, you know where that residual value is. Right now, we simply are too new in this market cycle, where reinsurance companies, insurance companies are studying this too, to know where that residual value guarantee is going to be priced at once that actually develops, once we have more price history. So on Bloomberg, you can see the old chips’ residual value index now. That’s obviously the first step. And you’re gonna see CME trading futures. So, all of that is gonna start creating more knowledge around residual value and what these things can be priced at if you want it to be an asset class on its own. 

Sanjeev Krishnan: If you’re an equity investor, you basically take a bet on tokenizing the service economy, which is twenty to thirty trillion. But on the debt side, you’re effectively taking a put option to hyperscalers that have tremendous EBITDA to cash flow conversion, and also you’re taking a bet on the equity investors staying there. But you have a lot less underwriting of the residual risk and the recontracting risk.

Karen Fang: So now obviously we’re being asked to finance data centers with neoclouds as tenants rather than hyperscalers. That’s where you start to see the market pricing start showing some inkling of, wait a second, what is this? Because obviously it’s not like a double-A‑, triple-A‑, or single-A-rated hyperscaler. So I think that’s where the market’s starting to learn and try to understand a little bit in terms of what this asset class actually means.

Sanjeev Krishnan: Well, I could talk to you forever about this, but I also want to get to one of our final points, and not to make it a footnote point, but actually a core point. You and Brian Moynihan have been leaders in sustainability and sustainable finance. How do you see the work you’re doing on the infrastructure and capital solution sides, and the link to sustainability, broadly defined?

Because I would define social license and sustainability together, which I think will be a core differentiator for firms to get permitted and built. I was taking a bit of a bet with someone yesterday that I think more gas pipelines to LNG will be permitted than gas pipelines to data centers, just because of the social license dynamic. And you’re having red states deeply concerned about this. This is not a partisan issue now. I’m just curious how you link some of the sustainability and social license side to all this financing and CapEx. 

Karen Fang: Yeah, so first of all, I would say I’m deeply motivated by sustainable finance, and that’s why it’s still part of my job, a really important job. Our $1.5 trillion sustainable finance goal from 2021 to 2030 is well underway. We actually just crossed over the trillion-dollar threshold. We’re about a year, year and a half ahead of pace. One of the reasons my infrastructure job versus my sustainable finance job isn’t separate is because the massive amount of demand for power for compute is driving a lot more renewable energy and storage deployment. Our power pipeline right now, if you look again, it’s not just our pipeline, but also look at the United States power grid, new capacity added in ​’25 and ​’26 is going to be 90% renewables and storage. And more solar than wind. So that’s just a proof point: it’s faster and cheaper. And it’s actually, frankly, especially given some of the IRA and OBBB tax incentives, it’s quite attractive from a project development standpoint.

So I’m happy as a sustainable finance person to be driving infrastructure. For me, it’s always about whether I can unlock excess power with batteries or build a solar-plus-storage system, because it’s faster and cheaper. The developers want it, and hyperscalers want it. Yes, there is gas demand because renewables are not firm. The batteries are making it more firm, but it’s not 100% ²⁴⁄₇ firm. So we still need advanced geothermal, which is also great because it’s lower-carbon.

You know, you saw we brought Fervo public, but gas plant demand is for sure there. 

But because of the turbine shortages that we talked about, only a small handful of projects are under construction rather than being planned. So, for me, I’m seeing the proof point of sustainable infrastructure being a big part of this digital infrastructure. And I’m happy to see that. Look, there are a lot of smart people at Bank of America. Brian Moynihan, our CEO, and Jim DeMare, our co-president, didn’t have to pick me to do all the infrastructure. Many, many competent people can do this job. I think they wanted me to do this job because we look at infrastructure across the board, and making sure a data center, if it can be a green, energy-efficient data center, that’s what we prefer. We always work with the developers and the tenants to actually look at technologies that can be implemented, whether it’s cooling and power and energy efficiency, and all of the above, to make sure that the holistic story is lower carbon, lower water usage, and lower power usage. 

Sanjeev Krishnan: I love the fact that you can operate at fifty thousand feet and one foot. What are you experiencing, maybe as a concluding point for our listeners day to day, that maybe will be in the headlines six or twelve months from now? You’re on the front lines of this, and you’re probably seeing things that other people aren’t. Is there something that you’re seeing that’s non-obvious from a headline perspective that you’re living right now?

Karen Fang: Well, I think it’s a great question. Last year, I would have said that six months from now, you’re going to see one of the largest data centers financed in Michigan, without producing new power but unlocking excess power with batteries, which we just talked about. That’s the project we financed in Michigan, in Saline Township, for OpenAI and Oracle that we completed in April of this year. That’s one of the largest senior first lien asset-secured, a little wonky, forget about the name, project bond that we designed, the 144A capital markets. That was a $14 billion debt financing for a $16 billion data center project. I use that as an example because I think Related Digital, the developer, and Blackstone, the equity investor, did a really good job engaging the communities in terms of understanding communities’ concerns with the fire department, the community center, and union jobs. I think there is a lot to be said about community engagement, given the current sentiment, because we are underbuilding, as we talked about. The oversupply risk is low. 

But where we have the time and the constraint to give us more time to think about where to build and how to build, I think we need to think about what I consider relatively good examples of the permitting, the power, don’t waste and don’t just build random new generation everywhere. Instead, think about how we can unlock excess power with batteries and use it more efficiently, engaging communities early and transparently. 

So that lends to the answer to your question, which is that we actually worked on a very large-scale chip financing in June. Now we’re leading as one of the very few banks leading one of the largest chip financings in the world. And the structure is evolving from even the time that we did the June structure, and the size is a lot bigger. So we’re learning a lot of lessons as we go on, how do investors price this, right? It’s ultimately not all going to be guaranteed by hyperscalers. 

How do we think about capital markets, asset allocation between investment grade, high yield treasury investors into this new asset class? How do we think about community engagement? Where are these chips going? How are they going to produce the compute? What is this compute used for? What is the social license that we all have to operate this massive amount of infrastructure that we’re building? 

So for me, it’s going to be this narrative of we don’t understand what this is going to do, this technology, and we have to build a lot of physical infrastructure for this. Is it worth it? Are we doing the right thing? Are we doing the wrong thing? And will people, who are the drivers of the destiny of this technology, accept or not accept? And what is our engagement education strategy? 

I do think the core element is the people. What are we doing for the people, with the people? How do we think about engaging communities and making sure this technology is going to help all of us win?

Sanjeev Krishnan: Well, I think on the people’s side, we’re very grateful that Brian chose you for this because you’re in the middle of some very important megatrends. And thank you again for joining us. I’m excited that our listeners could get an ounce of your wisdom here on this podcast. So thank you again. 

Karen Fang: Well, coming from the very wise man himself. Thank you so much, and thank you for flattering me. But it was a great conversation. I really enjoyed it.

Tonya Bakritzes: Thank you for tuning into the S2G podcast. If you enjoyed this episode, please subscribe, leave us a review, or share it with a friend. For more information or to connect with us, check out s2gin​vest​ments​.com.

This podcast is for informational purposes only and should not be construed as legal, business, tax, or investment advice, nor should it be used to evaluate any investment or security. It is not directed at, or intended for, any current or prospective investor in any investment vehicle sponsored by S2G. Investing involves risk, including the potential loss of capital.

References to specific companies are for illustrative and educational purposes only and do not represent an endorsement or investment recommendation.

The views expressed by podcast guests are their own and do not necessarily reflect the views of S2G. For additional important information, please visit: s2gin​vest​ments​.com/​d​i​s​c​l​o​sures.