From Lab to Scale: Tina Tosukhowong of TDK Ventures on What Makes Deep Tech Commercially Viable

Breakthrough science is only the beginning. The harder challenge is turning promising technology into a system that can operate reliably, economically, and repeatedly in the real world.

In this episode of Green Giants: Titans of Clean Energy, Wes Ashworth, host and President of Lee Group Search sits down with Tina Tosukhowong, Investment Director at TDK Ventures, to explore what separates deep-tech companies that scale from those that stall between the lab, pilot plant, and first commercial facility. Tina brings an unusually valuable perspective to venture investing. Before becoming an investor, she spent years as a chemical engineer and industrial operator working across process development, pilot operations, scale-up, commercialization, manufacturing, and customer qualification.  

That operating experience shapes how she evaluates climate tech, advanced materials, AI infrastructure, and other capital-intensive technologies today.

Tina explains why scaling is not simply a matter of building a larger version of a successful pilot. At commercial scale, entirely new risks emerge around heat transfer, fluid flow, contamination, materials handling, uptime, permitting, utilities, feedstocks, safety, supply chains, and integration with customer sites. The companies that succeed are often the ones that identify failure modes early, run the right experiments, and learn faster than their capital disappears. 

The conversation also tackles one of the most important questions in deep tech: When has a company actually earned the right to scale?

Wes and Tina explore why raising too much money too early can be dangerous, how founders should think about risk retirement before committing hundreds of millions of dollars to a first commercial plant, and why the strongest financing strategy aligns capital with specific learning milestones. They also discuss founder-market fit, coachability, executive hiring, productive board conflict, and why the right investor syndicate can matter as much as valuation. 

The discussion then moves into the rapidly evolving physical infrastructure behind AI. Tina breaks down opportunities across data-center power delivery, energy storage, cooling, thermal materials, networking, semiconductors, and photonics. She explains why the next wave of AI infrastructure innovation may come not only from better chips, but from everything required to power, connect, and cool them efficiently.  

The episode also explores fusion energy, synthetic fuels, critical minerals, and frontier technologies, including Tina’s perspective on Type One Energy, INERATEC, and the growing overlap between clean energy, computing, advanced materials, manufacturing, and national supply-chain resilience. 

In this episode, you’ll learn:

•  Why industrial scale-up exposes risks that laboratory success cannot predict 
•  What founders should prove before committing to a first commercial facility 
•  How capital discipline can prevent expensive first-of-a-kind mistakes 
•  Why learning speed and coachability matter so much in deep-tech leadership 
•  When experienced executives help startups and when they can slow them down 
•  Why investor syndicates and boards become critical when plans break 
•  Where Tina sees overlooked opportunities in AI data-center infrastructure 
•  Why cooling, rack-level power conversion, and photonics could become major bottlenecks 
•  What industrial operating experience reveals about the future of fusion 
•  How e-fuels, critical minerals, AI, and advanced materials are increasingly converging 

For founders, investors, operators, and leaders building the next generation of clean energy and deep-tech companies, this episode is a practical look at what it really takes to move from possibility to commercial reality.

Links:
Tina Tosukhowong on LinkedIn
TDK Ventures Website

Wes Ashworth: https://www.linkedin.com/in/weslgs/

• Email: wes@leegroupsearch.com
• https://leegroupsearch.com/green-giants-podcast/
• https://leegroupsearch.com/

Transcript

Copy Transcript Copied

Wes Ashworth (00:12)

Welcome back to Green Giants, Titans of Clean Energy. And yes, that's a slightly new name. It better reflects how the show has evolved to cover the full clean energy ecosystem. Today I'm joined by Tina Tosukhowong, Investment Director at TDK Ventures. Tina is a chemical engineer by training, but what makes her perspective especially interesting is that before becoming an investor, She spent years actually scaling industrial technology, working across process engineering, pilot operations, commercialization, and manufacturing. Today she invests across climate tech, advanced materials, and increasingly the physical infrastructure behind AI. So this conversation is really about one question: What separates technology that works in the lab from technology that actually survives the journey into the real world? With that, Tina, welcome to the show.

Tina Tosukhowong (00:57)

Thank you so much, Wes. Thanks for having me and I'm super excited to share my experience and having fun.

Wes Ashworth (01:03)

Yeah, it's a pleasure to have you. I'm really, really excited to get into it as well. And as we always do, we'll start a little bit with your journey. So as I said there, you spent years as a chemical engineer and startup operator before becoming an investor. What do you see in a deep tech company today that someone who has only ever invested might miss?

Tina Tosukhowong (01:21)

Yeah, I think that is a great question, yeah. Having been a chemical engineer and scaling technologies, the thing that was most challenging for me is like when you scale the first of a kind process. That's you know everything that you thought you might have known, you might have tested at a small scale, there's always unknowns that you encounter when the first of a kind plant gets built. So the real question that's always worries me is like if we build this ten times or a hundred times larger and run it for so much longer than what the setup data show me, can it really maintain the yield efficiency or performance number? And when they first run the pilot, sometimes they just have fresh solvent but it you have to recover this solvent a hundred time and the purity dropped twenty percent, can it still maintain what they show me on the data? So give you a good example. Recently I was looking at a startup company using chromatography for critical mineral separation. It was so elegant and I asked them the question like, How much water do you need to use to do this separation? And then I calculated they have to use like a hundred times the water to the feed stop ratio and I'm like, my, how do you actually scale it? So what I think the operator background really helped me was that when I had to think about really scaling this, deploying it in the field, it's not just about that particular process or reactor that they showed me. It's dealing with the interconnections of utility, permits and how you actually integrate this particular technology to the customer's site. It's a lot more complex and any mistake can be really costly. So it's not like a software investment that if you write a wrong code, there's a delete button that you can click and erase your mistake completely. This is deep tech. So I think the tribal knowledge and the experience is very important.

Wes Ashworth (03:31)

Yeah, I agree. I think that operator lens is so important. And I love the distinction because kind of that technical diligence and operating judgment are they're related, but they're definitely not the same thing. And I think you have a very valuable perspective when you've actually been an operator and seen it happen and you're thinking about those sort of things of like, okay, but will that scale? Will it commercialize? You know, does this have a kind of a long future ahead of it? So When you were helping scale industrial biotechnology, kind of different lens, what did the first sort of real plant teach you about the difference between technology that works and technology that really can become a business?

Tina Tosukhowong (04:04)

Yeah, that's such a great question. in my first real plant it taught me that the technology can't just work for a few batches. It has to work every single day 24/7. Otherwise it's not a business. And at plant scale, you also discover how impurities can affect your process, how different feedstock and how different components interact. So very quickly you discover that the uptime and consistency can matter so much more than the peak performance. And a commercial process it has to produce the right product at the right quality repeatedly. And the cost has to be something that the customer can accept. The process can become a business. it has to be really boring, really predictable. and make the same thing over and over again. And the error or the offset material you have to be able to reach like a few in the millions so that's within the six sigma to have a real commercial business for many many of the products.

Wes Ashworth (05:10)

Yeah, absolutely. I that sort of that gap between it works and it works repeatedly at commercial scale is where so much of the real story lives. And I think it's something people outside the industrial technology often maybe underestimate. You know, the business doesn't begin just because the science succeeds or so much else that has to go right. And you know, you know, that kind of boring stuff that really does make it scale. So I love that a lot. you know, again, I think this operating history gives us the lens, right? And now I want to get into the stage where deep tech really starts to become You know, expensive, physical, and unforgiving. And so, what is the biggest myth about scaling a climate or deep tech company?

Tina Tosukhowong (05:46)

Yeah. I think maybe the biggest myth is people may think it's just simply mean building a larger version of the pilot. But I think in reality, the physics such as how the fluid flow in the pipe or the reactor or the equipment control and supply chain or even permitting and financing, which is like other parameter that technology team might not be thinking about, that can change a lot as you scale. So the question is not only whether the core technology performs, it is whether you put the entire system together, can it still perform reliably under that realistic operating condition? So scaling is it's a learning process and I think the companies with experienced teams that can identify and anticipate the uncertainties that matter the most. they can be really well prepared and design the experiments to solve them early. And that learning rate I think it's very, very difficult, very important in order to move quickly and make it meaningful and scalable and reduce risk as you try to scale the process.

Wes Ashworth (06:59)

Absolutely. That's so good. And so with that in mind, kind of thinking about that, how does the founder know the company has actually earned the right to build the next scale of plan?

Tina Tosukhowong (07:09)

Yeah, I think that's a real million dollar question, right? Multiple million dollars question.

Yeah. I would say that there's no one magic milestone, and I always look for body of evidence. I think the company really need to understand their operational envelopes, yield, throughput, uptime, qualities, And responses to the variability in all these parameters very early on. And they should test those components that's most likely to fail and have a credible plan for control safety, supply chain, and permitting. But it's not just that. it's also the market has to be derisked look for the evidence of the customer traction.

I look for the evidence that they indeed produce the product that has the right quality, have been qualified and there's some evidence that they can do that consistently, not just showing me the golden sample. And also would the customer pay for it? oftentimes in the last clean tech cycle we offer like green product, but the customer they liked it. If the company reached the price parity, They'll use the green as a tiebreaker, but when you ask them to pay premium, they often like shy away unless there's a real mandate or something. so I think that both the technical and commercial has to be real evidence to support that they are ready to scale and then that also unlocks the financing for the next phase that will help them scale the technology.

Wes Ashworth (08:45)

Yes, it's such great wisdom there and a great way to think about it. And I really like that idea of scale being something a company sort of earns rather than something capital simply buys.

Wes Ashworth (08:54)

so I'll ask you this question, me and it's kind of directly tied to that. So if a company is about to raise 200 million for a commercial facility, what would you rather see it spend 10 million proving first?

Tina Tosukhowong (09:06)

yeah. I definitely would like them to spend ten million proving the failure modes of the future plant before they build a larger version of the plant and running this critical sub system for a much longer time, than the lab has done so that we don't have to worry about what it will look like if we fail on the field because if you're gonna spend two hundred million to build a commercial facility, either you hit it out of the ballpark and unlock, you know, the unicorn status or the company can run into a fatal mistake that they just have to shut it down and become a distressed asset.

Wes Ashworth (09:47)

Yeah, absolutely. And it, you know, it's where capital discipline gets very tangible because a smaller experiment can sometimes you know prevent a much larger mistake, is what you're sharing there. And you know, every dollar should ideally sort of buy more than equipment. It should buy, well, you're hitting on it there's knowledge, you know, knowledge of what's gonna fail, how it's gonna fail, how do we prevent that? And prevent that to where it's a much larger scale with which much higher risk as well, too. So I really like that thought a lot. so here's an uncomfortable question, maybe.

Can raising too much money too early make a deep company worse because it lets the company scale faster than it learns?

Tina Tosukhowong (10:20)

Yes, I think absolutely. And I've seen a number of examples in Clean Tech 1.0 there's no company is too big to fail and the issue is not about the money itself, and it is whether the capital creates discipline or remove it. some people use the capital in the wrong way. building before the technical risk has been retired or hiring. and building such a huge commercial team before they understand the product market fit. So overbuilding too early and committed to design when things have not been adequately tested, I think that can be a fatal mistake that startup can make in the deep tech system. But having said that, racing not enough is a bigger problem as well, right? it can be equally damaging. So I think the right approach is really to align the capital with the milestone and learning objectives and using the financing insights where it is possible. And you kind of show the investor as you go that you have to retire the risk and therefore have the solid plan and not dilute the equity investor too much. by the time we go to commercial.

Wes Ashworth (11:37)

Yeah, absolutely. So good there. I love that you made both of those points and kind of looking at both perspectives, right? I think it's about that finding that balance and that sweet spot. And you know, I think overall it's a counterintuitive point because fundraising success and company progress can look identical from the outside when they're really not. so I love that. And hey, we're trying to prevent these fatal mistakes, right? so once the money is wired and the equipment finally turns on, where do the problems usually show up that the model maybe never captured?

Tina Tosukhowong (12:02)

Yeah, I think that's a good question. there's a smart person who's taught that all models are wrong but some are useful. And what I often see is that models are generally very good at predicting the average. But the planned operations it has a lot of variabilities and failure modes. And I've seen very commonly if you're talking about inside process technologies the model may predict the reactor is run at, you know, a hundred and fifty degrees, but when you have such a large piece of equipment, the heat management, the fluid flow, the mouth distribution becomes so common. You may have hot spots in the reactor and that high temperature generates side reaction. And you may have material handling like clocking that show up that we didn't build it in the operation downtime because we didn't anticipate. Or contamination is very common in biotic processes and things like that. So all these I need to mention safety issue as well. It always show up when things get larger. So model may assume a certain plan up time but in reality it all depends on the learning rate of the operation team. and how fast they can learn and really ramp up the production and uncover the new problem and debottleneck the whole plant. And I often find that it's relatively simple, if you go from zero to sixty percent of the plant capacity. But after that every ten percent improvement it gets so much harder. So I think that's why experienced team who pay attention to detail and with the battle scar I think it matters enormously.

Wes Ashworth (13:47)

Yeah, a lot to be you know, unpacked there. And but I love multiple times you've come back to sort of the people side of it, right? It like how quickly they learn and adapt and the operating team and those just how important that is to have those right folks in place from the beginning. so I love that a lot. I love that perspective. so what is maybe a milestone that makes you stop saying, promising technology and start saying this can actually scale.

Tina Tosukhowong (14:10)

Yes. I think for me it ties back to whether the company really shows that they can take the first principle and prove that it can become reality. I think the first principle is important because it allows you to predict beyond what you operate currently. When you put in a new variability it can give the prediction based on the specifics, based on thermodynamics, right? But the hard part is always matching the model to reality. And recently I actually have one portfolio company. It's a German company called INERATEC. They produce electro fuel from converting carbon dioxide and green hydrogen into sustainable aviation fuel. And they have built a first of a kind facility just close to Frankfurt Airport. And they show recently the capacity ran up to steady state within like twelve hours and they show a real time digital twin prediction and the actual material and energy consumption and the number that they show me is like within plus or minus eight percent. And I was like, wow, this is what I expect to see in actual petrochemical refineries and I see it in a startup company and I thought, Wow, this could actually scale. So and what I'm very proud of recently a commercial airline KLN has also flown from Amsterdam to Hamburg with five percent blend of INERATEC fuel. So I'm very excited if they can show that they can do this like eight thousand hours out of the year, I think we're gonna have an e-fuel industry that prosper and really reduce the fossil fuel. So I'm super excited and I'm looking forward to seeing this kind of example in many more technologies in clean tech and deep tech industry.

Wes Ashworth (16:07)

Absolutely. And I appreciate you sort of making that real and the real example with INERATEC and you know, as you mentioned, so a company turning CO2 and renewable hydrogen into synthetic fuels and they have gone from technology development to actual commercial fuel production is incredible. anything else sort of like what happened along that journey that maybe people outside the company really would never see?

Tina Tosukhowong (16:26)

Yeah, I think the big story was that they are converting CO2 and hydrogen to fuel but I think the hard part is like the amount of integration required to make that possible. It's really, really complex. It's this is the technologies that you have to convert CO2 and hydrogen to intermediate. one reactor, it consume a lot of energy. Another it releases a lot of energy and integrating this and optimizing the energy flow, material flow is super complicated. And petrochemical industry, you know, they have eighty years to learn this. And startup they have such a short windows that it's so complex to demonstrate this. So the journey that the founder has to take I think it's very heroic and the amount of blood, sweat and tear that they have to put in working, you know, twelve, hours, to get this turned on and to prove to investor that this thing can scale is enormous. So I think we, as the investor I think we get the easy part, right? Get to evaluate them to judge them. But the people behind that state I think they are the one that I consider the hero of the clean tech story because inventing something new against the incumbent is so hard.

Wes Ashworth (17:50)

Yeah, so difficult. And this is why I love using a real example is that milestone often gets the headline, but it's those years of just the problem solving behind it that usually don't. And it's that grit, sweat, and blood and tears that go into that. And you know, there's big difference between, you know, reading that, hey, a plant opened and understanding really what it took to get to that point and get the plant operating. so kind of just further on this, so at what point should a climate tech founder stop thinking like a venture back technology company and really start thinking like a project developer.

Tina Tosukhowong (18:18)

Yeah, I think the transition should begin earlier than many founder expected. I think around series A I think that's a good time to start thinking about it. If you are thinking about building the pilot for series A, you should also as much as possible think about what a larger scale project should look like and what kind of risks you have to encounter and try to bake that in as much as possible. So because I think different sites that they go in, right? It may include different feedstock. So they need to make sure that they consider the feedstock they would have utilities and even the product market fit, right? Maybe the process generates multiple types of product, but you have to kind of like know that who your customer's gonna be. So you need to start lining up. those partners with utility, land, feed stock, off taker, EPC, and well in advance because the time is limited in the as a venture back company. So I think before going to series B you definitely have to have the evidence that You know how to build projects so you can think about deployment and things like that. Have to try to keep in mind the requirements.

Wes Ashworth (19:39)

Absolutely. Such good advice there. And I think, you know, that transition really does seem critical because as you're saying the sources of risk, sources of capital start changing at the same time. So get to start thinking like that early. so as we've talked about, you know, scale and we've mentioned some other things. So scale is isn't the only an engineering problem. It's the market, the people around the table, the structure of the capital can determine whether the company survives the inevitable detours. And so You put market size near the top of the list because a small market is one of the few things a great team can't fix. How do you tell the difference between just a genuinely huge market and a pitch deck that simply, you know, found a huge TAM number?

Tina Tosukhowong (20:16)

Yeah, for me, actually TAM is not as important as the growth rate. If the market has a huge TAM but it isn't growing, it's really hard for newcomers to replace the incumbents because the incumbents they have a lot more resources, manpower and market channel. So the market has to grow. There has to be the growth driver, otherwise very hard for startups to really penetrate. And I think another thing that venture investor and founder need to consider is the sales cycle. Even though it's big with the lush CAGR what does the sales cycle look like? venture funds typically have like ten years of fund life. If the sales cycle for a single piece of equipment is like six to eight years, it can be really problematic. So To understand the demand sales cycle, I typically work from the customer backwards, not from the market report forward. I have to try to understand who is the buyer, what's the problem they're trying to solve and what their decision looks like, what's the budget look like. Because it's not just the end product, like can someone actually pilot the technology with the startup while they are still iterating the technology. So I think those came into play a lot when you try to think about the market it's not just the size, it's the growth and also the sales cycle that matters.

Wes Ashworth (21:40)

Yeah, I love it looking at that kind of full equation from your perspective and understanding that. And you know, I think that distinction really matters because you know, almost every startup can find a way to produce an impressive, you know, market-sized slide, but there's a lot more that goes into it, a lot more that you're evaluating. So I think that's really important. so most investors say they back great founders, just as a general rule. I know you focus more specifically on just founder market fit and learning speed that we've talked about throughout today as well.

What are the clues that tell you someone is truly coachable rather than just simply good at fundraising?

Tina Tosukhowong (22:09)

Yeah, I think that's a good question and sometimes it's not apparent when they pitch to me, right? but when I think about it, I think a coachable founder is someone who is willing to update their point of view when the evidence changes. And I look at how they discuss like bad news, failed experiments or maybe the lost contract. and things that didn't go well, I often ask them like what was their biggest failure in the past and what have you learned that have changed your thinking. and some people they immediately try to jump and blame someone. so I think you can tell a lot by maybe spending a little bit more time with the founders to understand their mode of thinking, how they approach a problem, how they solve problems, and are they receptive of recommendation and willing to change their views. So I think the strongest founders they listen carefully, they make decisions and they then they act decisively.

Wes Ashworth (23:13)

That it's so good. I love that perspective so much. And I think, you know, learning speed feels especially important in deep tech because the company is gonna encounter problems nobody has seen before and you have to just kind of roll with it. And so you can't have this huge ego attached and be hyper defensive and be willing to change and be nimble. but I love that is a lot. I'd love to love focus on people. I think so many times people just focus on technology or, you know, any number of things that we've talked about, but just really this coachability, you know, and how important that is so cool to hear. why do so seasoned executives from maybe mature industrial companies sometimes just really struggle inside deep tech startups?

Tina Tosukhowong (23:50)

Yeah, I mean I've seen this played out very often, right? And I think there are several I don't want to label but I think I've seen seasoned executives from very large companies, if they were part of that journey, from small players to become large, like let's say Tesla, like they tend to understand what it takes to be in the growth. phase growing from you know zero to something. But if they come from the mature industry that's already have everything in place, processes, resources, brand credibility, oftentimes they could get lost in the setup that may have none of those in the early days. And I think in early stage D Pick setups, Founders may have conceptualized new technologies to disrupt incumbents and they have to figure out product market fit, right? And I think that's the most difficult space because there are multiple ways you can grow the company which you just need to pick the right path. And oftentimes I find it difficult to bring seasoned executives from mature company into that space before they have the product market fit. but once the company has found their product market fit and it's ready to grow, I think having the seasoned executive to grow the company from ten to a hundred, I think that's a much better stage when we bring them in.

Wes Ashworth (25:14)

Yeah, that it's really a fascinating kind of hiring tension. cause I think experience can be incredibly valuable while still being optimized for a completely different environment. So as you said too, some of it's timing, right? It's like you're not ready to bring that person in just yet. You

Wes Ashworth (25:27)

know, wait a little longer until that product market fit is there and what have you. So we I obviously get to see that play out in real life all the time, you know, when it's done well and right and when it's not sometimes as well. but yeah, I think that fits incredibly important all the way around. so how do you know? So when a technical founder should so remain a CEO and when they really need a strong operator beside them, and when the company really needs a different CEO, how do you kind of like figure that out and strike that tension?

Tina Tosukhowong (25:54)

Yeah, that's a very important strategic question that every company with technical founders, CEO, we have to go through this process and I think the technical founders should remain the CEO when they can do more than just lead the technical vision. they need to be able to attract talent, they need to be able to understand the customer and communicate clear vision. and make decisions and most importantly build a strong organization around themselves. If they tend to be a one man company, it's about me. I think that's very worrisome. Yeah. And but sometimes I see the technical founders very versatile, very adaptable and I tend to like to surround the CEO with other seasoned technical experts to bring in different capabilities while letting the CEO carry on their vision and communicate those visions to investor and stakeholders. yeah, and then we touched upon these and when the company stage has changed then maybe the founder themselves may not want to change with it or are unable to change with it. And the decision to change the CEO I think should be based upon like the company's needs and be handled with respect. And we don't want to diminish the original founders but sometimes we just need to make the right decision for the next phase. And I think the transition is best when the founder has a strong self awareness and work together with the board. to find the right successor to pursue.

Wes Ashworth (27:40)

Yeah, these incredibly consequential decisions. Cause that, you know, the right leader for one stage, as you said, isn't automatically the right leader for the next. And I think be having that awareness, having the humility, the openness kind of to see it as well. You know, I've worked with some of those kind of early stage ones where they really do realize it and have that self awareness, like, hey, I really need somebody, you know, that's an operator that has scaled to come beside me, whatever that need may be to balance, you know, kind of where they are. So I think that's such a huge part of it, the human side, that maybe isn't talked about as much. So I don't want to get that out there as well. so you mentioned boards and we'll transition a little bit there. So the best boards are, you know, not always the most comfortable boards, and I would say probably the best boards are not the most comfortable boards. But what does productive conflict look like when the company is under real pressure?

Tina Tosukhowong (28:23)

Yeah, I think the productive board is direct, evidence based and focused on decisions, instead of like trying to win argument or pointing fingers. And I think the one that I like, is the one that actually surfaces hard truths, and not just talking about win, win, win, but actually try to think about the growth mindset. What could we do better? what are the options. What do we know? What do we not know? what happens if we take certain decisions or we wait. And I think healthy board discussion would allow this kind of like discussion and disagreement before decision. And once decision is made then everybody align and rally to support the management to execute. And I think that's when magic happens, and when companies can take Decisive decision and move forward.

Wes Ashworth (29:15)

Yeah, absolutely. and I think that word productive, right? And productive conflict is it matters because disagreement only helps when everyone still trusts that they're working towards the same outcome. But how just important and necessary that is in that sort of conflict, sometimes in being uncomfortable and hearing something different, you know, silence can look like alignment, right? Up until the problem becomes too expensive to ignore. So how important that is to have that sort of healthy board environment and productive sort of level conflict working towards the same solution or an outcome that everybody's everybody wins when we get there, but you know, it may be a little uncomfortable as we go. so as the company is going, so when the plan breaks, which we've talked about this sort of throughout it, this flexibility and kind of learning as you go and inevitably plan usually does break at some point. what does a great investor actually do? Maybe that a the passive investor doesn't

Tina Tosukhowong (30:03)

Yeah, I think a great investor don't wait for the next board meeting to hear bad news, right? We really call up the company and figure out what's going on. is this a temporary setback or there's a fundamental problem and then bring the relevant expertise and relationship to help the company navigate the implication that event might have cost. It could be like rich funding, it might be like we really need to hire the right experts to really fix this problems. So I think this really means a lot. when we see someone roll up their sleeves. it means a lot in deep tech because it's so difficult to build a company to become successful. So rolling up sleeves at the right time it goes a long way.

Wes Ashworth (30:54)

Yeah, I like that a lot. It's probably the moment kind of founders discover really what kind of investor they actually raised money from. And, you know, it's easy to be supportive about everything's working. There's harder moments to just expose the quality of the partnership and seeing, you know, how meaningful that can be, sort of like you said, rolling up your sleeves at the right time. so founders often optimize for the highest valuation. In deep tech, should they really be optimizing for the strongest sort of syndicate instead? Like what are your thoughts on that?

Tina Tosukhowong (31:18)

Yeah, I think for me I believe the syndicate is at least as important as the valuation because the right syndicate can bring industrial expertise or customer access or manufacturing relationships. and also some may have special relationship with project finance or regulatory framework. I think those are very important. And not to forget ability to follow on. That's super important because in some of the deep tech it takes a lot of capital to really scale the technologies. So having a continuity in funding is really crucial because I think if the lead investor don't follow it could be a pretty bad signal or for the companies. But I guess it's not it doesn't mean that the valuation is irrelevant. The founder should understand the delusions and try to preserve a healthy capital structure. But I don't want that to get in the way of finding the right long term investor that could help you for the long run and just stuff with just the number of the valuations.

Wes Ashworth (32:24)

Absolutely. I love it. Love it. It's a different way to think about fundraising. I think that the best round may not be the one with the biggest headlines sometimes and it's just how it all comes together and how important that is. so as we've talked about this, you know, I think these same lessons around infrastructure, bottlenecks, and physical constraints are now showing up in another, you know, massive market. So a AI is forcing the energy, materials, semiconductor, and competing worlds into the same conversation. And so you've said, you know, AI infrastructure is becoming impossible to ignore. And I would agree, because the challenge is not only generating more power, but using that power more efficiently. from where you sit, like what are the biggest efficiency opportunities inside the data center right now?

Tina Tosukhowong (33:02)

this can be a long topic that we could talk for hours. Yeah. at my firm we look at this as three important arc. the first are is power. Today data center draws a lot of power from the grid. They're becoming a big, big user. And if you look at the power from the grid hundreds of thousands of volts and it has to go through several energy conversion and by the time it gets to the chip it's less than a volt and each of these power conversion it's incurs a lot of power loss. So data center needs to they need to look for energy efficiency and energy reliability as well. And we invested in a few companies in this space. one is peak energy, a great scale, sodium ion energy storage company. for data power for data center power reliability. And they have a much higher round trip efficiency compared to lithium ion batteries. So that's super important for the grid. And another one is Amperesand They are developing a solid state transformer that can achieve very high power conversion efficiency from medium voltage to direct level voltage. And then I think the third one we actually had invested in C2i Semiconductors, they do a power delivery to the chip scale. we announced the investment a couple of months ago and last week. Infineon announced the acquisition of the company. So it's really an important space. And then I think a second part is the cooling, because any energy that's not delivered to the chip, that goes into heat. So you need to look for highly energy efficiency cooling mechanism for the data center. And we invested in Fabric8Labs which does the liquid cooling coal plate and that would reduce the energy consumption in the data center. And then NovoLINC is another one. They make thermal interface material that lies between the chip and the coal plate. To move the power to the heat sink more efficiently and improve the reliability of the chip. And finally, we have to talk about increasing the utilization efficiency of the compute asset. GPU is super expensive, but Meta did a study and showed that 30%, 40% of the time the GPU is sitting idle, waiting for the data from the network. So we need to improve the speed, the bandwidth, and reducing the energy consumption of the network equipment. And we have invested in Mixx Technologies. they are bringing photonics to the AI networking within the data center rack. And we also invested in Eridu AI that develops the next generation heat of switches that provide connect. than with improvement for data center. So a lot of innovation around the data center because I think it's gonna become one of the most important load user in the future.

Wes Ashworth (36:06)

Yeah, I would agree. And I appreciate kind of going through just even those investments you've made too. And it really does paint the picture of just that holistic view of all the different components and all the different pieces, all the different levers we can pull there and just how important that is. I know you've mentioned this as well, just in your answer there, but you as you've been looking at sort of this cooling, networking, rack level, power use, which of those bottlenecks do you think the market is still underestimating most?

Tina Tosukhowong (36:29)

Yeah, I think judging from a lot of investment that's going into networking and many startups have extremely high valuation. I think the underappreciated investment areas probably cooling and also the rack power conversion, I think those are underappreciated and I think we need a new approach because it's such a pick and shovel technology that No matter how many data centers you need, you're gonna need them.

Wes Ashworth (36:57)

Absolutely incredibly important. I wanted to make sure we put that in there so people can start paying attention to those areas too. so you mentioned it briefly there. So you've talked about the shift from the century of electrons toward the century of photons. what does that really mean in practical terms and why are photonics becoming more important as AI infrastructure scales?

Tina Tosukhowong (37:15)

Yeah, I think for the past century we use a lot of power for electronics and the electronics industry has gives so much wonderful devices that we use today. But photons travel so much faster and our understanding of photonics is improving. So I See that it's becoming more attractive to shift toward photonic communication to be able to move information across chips, boards and racks. And as I mentioned previously, optical communication, it can provide high bandwidth, low latency over distances and potentially lower the energy consumption per bit, then moving the same information using electrons. And I see that as AI workloads grow, the cost of communication becomes a larger part of the system. So I see that we need to move from electronic to photonic communication. But for sure, the reason why photonic is behind because it's also very challenging. the challenge in packaging, manufacturing, reliability and integration is quite a big challenge. So I think there's a lot of innovation and investment opportunities and I'm super excited what the century of photons can bring. I'm not gonna say the electrons are moving away, but I think we will see the more importance of photonic in many industries going forward.

Wes Ashworth (38:45)

Yeah, absolutely. I think, you know, a great example of why it belongs in an energy conversation too, and why it's exciting to look at this and sort of future innovation in that space and the investment that's happening in that space as well. So as we talk through a lot, I want to finish by just looking beyond today's constraints and into the technologies that could reshape entire industries and thinking about that sort of next frontier. So you've also invested in fusion through Type One Energy. having actually worked around industrial plants, what separates a sort of beautiful physics experiment from something you can really imagine sort of a utility operating for forty years?

Tina Tosukhowong (39:19)

Yeah, I mean I would say that physics is the sort of like admission ticket where you have the right to play if the physics work. But the more important thing to create a long lasting industry that can replace fossil is the ability to have a fission technology that can run continuously twenty four seven, just like how the coal fire power plants or gas fire power plant has run in the past. So I brought that twenty years experience that I have worked in the utility fossil power plant to the future investment. we invested in Type One Energy for two reasons. number one, the science is based upon Stellarator and Stellarator is the top four performing fusion reactor that has been innovated and improve over so many years. So we don't like taking physics risk when it comes to venture investment. We would rather take engineering risks. And we think that the physics of accelerator has been de-risked by a lot over the past seven decades. But second most important thing is that among the top four fusion concepts, accelerator is the only one that is running at steady state. And with my background working in power plants, typically you see the plant get turned on, operators can walk away and let the control overtake it. And it shouldn't take like PhD engineer or scientist to run the plant. It should be like something that blue collar worker can operate. And because it's inherently running at steady state, I think that this will pass the test of time, being something that's can run reliable, not putting too much stress on the material of construction and have a potential to finally replace the base load coal firepower.

Wes Ashworth (41:16)

Absolutely great perspective there. And I think, you know, once you think like a plant operator, reliability, maintainability, component life, fuel economics all become, you know, really part of the fusion question. So really great perspective there. Another main piece of this is just we've talked about this a lot lately on the show, is just critical minerals. they used to sound primarily like an energy transition issue. Now they're an AI, manufacturing, defense, national security issue too. Has that changed how you think about the investment opportunity when you're looking at different things?

Tina Tosukhowong (41:42)

Yeah, I think it has broadened my investment thesis significantly. I think back 2021, 2022 when people talk about critical minerals they thought about like battery materials, right? But I think there are so many more critical minerals that are important for energy transition and also the supply chain resilience of countries. Things like copper, rare earth magnets. you need copper for any application that touch electrons in data center and so on. You need rare earth for every moving cars and also the hard disk drive for data centers. So the problem is like US or European countries they don't have a lot of domestic supplies of those critical minerals. So the investment opportunities for us we have brought in the thesis. It involved like better extraction technology, refining, recycling, recovery from waste and efficient use of scarce input. So I think it is important area that we will continue to put a focus on because as we know, resources is depleting while demand for these minerals is like going up so this is an urgent issue that needs to be addressed.

Wes Ashworth (43:03)

Yeah, very urgent. You know, and I think as you just point out, there's the categories become much more broader than batteries and renewable energy alone and supply chain resilience as well. It just increasingly feels like technology strategy and not simply a procurement issue. Like there's a lot to it. so as we get into this, we're kind of like final questions here. And one of the things that I found really fascinating about your work is how frontier technologies just cross industries, you know, fusion and accelerators can influence.

Isotope production and photonics can reshape computing and materials can unlock new energy systems. And I guess where do you see maybe the most surprising crossover coming next?

Tina Tosukhowong (43:38)

yeah, I think it's very difficult questions, because I think that's always it always happens, right? and it's often non obvious at the beginning, when one technology is often been invented for something like tech fusion, right? it when people want to be successful at fusion energy generations, they look at particle accelerator, look at better lasers, look at like so many new material development. And then comes this same technology can allow you to make new isotopes for cancer treatment or new semiconductors. So all this crossover is really surprising and at the same time it is why we are keeping investing because when you invest in one technologies it open up the horizon and allows you to be able to invest in other because of the knowledge that we have learned. And I think with AI's proliferations, people may think it's all the software, but I think it's being crossed over into the physical space. I don't know how many pitch deck I received from startups each month. pitching a new AI tool to discover new materials and I think it's not just material, I think it's making the process technologies moving faster. So the development cycle is only gonna be faster with such a high learning rate that we see from AI's transformation. I'm super excited about the future and what that might bring.

Wes Ashworth (45:15)

Yeah, it's extremely exciting. I think those crossovers are fascinating. And it and more than anything, I think it's a good reminder that the biggest impact of a breakthrough isn't always the first application and that ability to learn, to be open, to pay attention, to see what, you know, people are asking sometimes. And I've had that happen too. You know, other companies that reach out to a founder like, hey, we were thinking about your technology, maybe can help us here. you know, and really kind of looking at those as well too. So really fascinating to watch that unfold. so final question here. You that does I think the common thread through all of this is the move from possibility to reality. And I want to close with the lesson you'd hand to the next generation of builders. So for the founder listening who is about to move from lab to pilot to first commercial plant, what's the one thing you would want them to understand before they scale? And then what makes you optimistic that this generation of deep tech can get it right?

Tina Tosukhowong (46:04)

Yeah, I think the most important thing for founder is that you do as much experiments you can and learn about the process as much as you can at the smallest scale experiments you can possibly do to test your hypothesis and refine your model. And also try to incorporate the customer feedback early on in your design and experiments. Sometimes you don't have to be perfect, You just have to keep learning and iterate fast because the world changed fast. The customer requirements are also changing fast as well. So you want to make sure that you incorporate as much input into your feedback learning route as possible. And then your other question, I think the what I'm optimistic about is that I think the future of deep tech is that we have a much better access to information and tools and I think that this would help the flywheel of innovation go so much faster and hopefully a lot of mistakes that the older generation has made the story gets retold and then there's a crossover of the knowledge from the previous generation to the newer generation so that we can innovate so much faster And hopefully some of this will translate to better economic growth, better treatment for disease and a more sustainable future.

Wes Ashworth (47:27)

I love it. That's so good and a perfect place to leave it. A lot to be optimistic about, especially when the science, the capital, the operating lessons start coming together. And so, Tina, this was absolutely fantastic. I really appreciate the way you connected the science, the people, and the capital and the operating reality behind all of this. there's so much excitement around breakthrough technology right now. And I think the bigger lesson is that breakthroughs only matter when we learn, how to turn these into systems the real world can actually use. To everyone out there listening, thank you for spending some time with us today. If you enjoyed this episode, make sure you subscribe to Green Giants Titans of Clean Energy so you don't miss what's coming next. Leave us a rating and review. And you can also follow us on social for clips, highlights, and more conversations from across clean energy, climate tech, and the broader energy transition. With that, we'll see you next time on Green Giants.