Pod Insights a16z Podcast 2024 Big Ideas: Developer Influence, Maritime Upgrades, and New Video Intelligence About this episode A new age of maritime exploration, the rise of the developer as a buyer in financial services, and more applications for computer vision and video intelligence. We asked over 40 partners across a16z to preview one big idea they believe will drive innovation in 2024.
Here in our 3-part series, you’ll hear directly from partners across all our verticals, as we dive even more deeply into these ideas. What’s the why now? Who is already building in these spaces? What opportunities and challenges are on the horizon? And how can you get involved?
Transcript Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.
0:00 Speaker 00 This has dramatic implications, not only on how financial services are built, but then also the services that we receive as consumers and businesses.
0:08 Speaker 01 How much of that data is actually being captured and utilized in some interesting way?
0:13 Speaker 02 Maritime in many ways has been the genesis of a lot of pieces of technology that have fundamentally changed the world.
0:20 Speaker 00 You can imagine a very low maintenance cost of financial services, which will open access to many more people.
0:26 Speaker 01 Where else can AI be applied that isn't what we've been seeing with all the exciting, generative, creative stuff and thinking about how these technologies can actually work in these more brick and mortar, physical use cases?
0:37 Speaker 02 Very similar to Amazon, you can't boil the whole ocean at once. You need to thread the needle and go from books to the everything store.
0:45 Steph Smith? Interpretable large language models, a new age of maritime exploration, democratizing miracle drugs, and anime hitting the mainstream. By now, hopefully, you know that we asked investment partners from across A16Z to preview one big idea that they believe will spur innovation in 2024. And our team did compile a list of 40-plus builder-worthy pursuits that you can find at a16z.com/megideas2024. Today, we close out our three-part series around these big ideas, together with three more partners from across our verticals. In Parts One and Two, we explored major trends in biohealth, consumer, and games, but today, we shine a light on the future of American dynamism and fintech.
1:30 Steph Smith? So who's the newest buyer in financial services? And how can unlocks in artificial intelligence change the value of unstructured video data? Plus, how on earth, pun intended, is it that we know more about the surface of Mars than our own seabed? Let's dive in. Pun intended again.
1:51 Steph Smith? As a reminder, the content here is for informational purposes only, should not be taken as legal, business, tax, or investment advice, or be used to evaluate any investment or security, and is not directed at any investors or potential investors in any A16Z fund. Please note that A16Z and its affiliates may also maintain investments in the companies discussed in this podcast. For more details, including a link to our investments, please see a16z.com/disclosures.
2:20 Steph Smith? First up, as software continues to eat the world, it also shapes the ultimate buyers of our products. As the industry has moved from, quote, marble to code, A16Z general partner Angela Strange believes we're seeing the developer rise in influence, ultimately laddering up to the kinds of products that are made.
2:39 Speaker 00 I'm Angela Strange, a general partner on the A16Z fintech team. And I think in 2024 will be a significant rise of the developer as the buyer in financial services. Historically, financial services infrastructure purchases were mostly driven by the economic buyer. What's my ROI? Or the business lead, does this solve my use case? But there is now a third increasingly influential constituent, the developer. Look no further than Move's growing Fintech Developer Conference and associated 4,000-strong community. In tandem, developers at larger financial institutions and insurance companies are becoming more influential in buying decisions. The rise of a developer as buyer in financial services
3:21 Speaker 00 companies of all sizes favors new entrants. For fintech companies that pride themselves on a great developer experience, this will play to their advantage. Fintechs are already prioritizing the creation of developer sandboxes to let customers try before you buy and are even open sourcing parts of their solutions. The developer buyer would, of course, prefer to get a sense of how the product works prior to implementation. But isn't that strategy also better for everyone? For large financial institutions selling their infrastructure, appealing to developer will be a new muscle that may require improvements in product architecture, including up-to-date documentation. Still, it's clear these institutions are recognizing the need to influence the developer as well.
4:04 Steph Smith? All right, Angela, I watched your talk from Fintech DevCon a few years ago that documented this multi-generational shift from marble to code. I love that line. And the ability also for almost any company to participate in this financial ecosystem, not just banks. I think it'd be great if you could touch on that history to really set the stage for this important new buyer being the developer.
4:26 Speaker 00 If you ask most people who has influence in the $25 trillion financial services industry, they'd probably say the finance leads or the lines of business leads. And what's happening now is there's a third very influential constituent, which is the developer. And this has dramatic implications, not only on how financial services are built, but then also the services that we receive as consumers and businesses. And I think anytime you talk about a big shift like this, it's sometimes helpful to walk through super briefly, like what have been the big eras, just to underscore the importance. And this is where the marble, the code sort of framing came from. And so you think not that long ago, banks were built out of marble
5:11 Speaker 00 and they were only concerned with keeping physical money safe. Fast forward to recent history, the mainframe arrives. Banks obviously have to keep virtual money safe. Those mainframes are still where 99% of dollars in the US are settled. So the trend that we've been talking about over the last several years has been the shift to as a service. And so it's taking some of the elements of the banking stack and making them such that financial services moves from something that traditionally only banks could do to now something that any company can do. Examples, Lyft can add bank accounts for its drivers or Procore can add payments and bank accounts for their plumbers. And what has happened is any developer at one of these companies,
5:52 Speaker 00 and even at the banks, has discovered that even though we have these as a service layers, there's still a lot of stitching that needs to go on to put these things together. And then to say the obvious developers are very smart. So anytime they find themselves building something over and over again, they will find a way to automate that and oftentimes open source it. And so we're starting to see all of these very well-used prolific, open source libraries that are solving what used to be esoteric problems, but that are used by millions of developers all around the world. And so this fourth shift is taking the as a service layer and exploding it even further such that you can think about it as almost having Lego for banking.
6:37 Speaker 00 So it's these different composable blocks that all fit together in different kinds of ways. And then the added bonus is many of those are open source. And so the smartest people from all around the world can contribute to these, making them even stronger and even better. So what is that new? You can build way better products, much faster. And it's starting to attract some of the smartest engineers are now working on financial services products.
7:03 Steph Smith? I love that. And I think maybe it'd be helpful for the listeners if you could give a couple examples of those financial services, infrastructure products that you say were previously driven by the economic buyer or the business lead that now you think are really being influenced by these developers.
7:19 Speaker 00 Yeah, exactly. So many vertical SaaS companies, for instance, so if you're providing, I'll use the Prokhor example, if you're providing a operating system for any type of business, it's a very natural extension that you might want to add payments. And traditionally, payments will expand the dollars that you are able to get from your customers. It's going to give you more margin. It's going to make people stickier. It was a very economic decision that this is the right thing to do. And so let's pick the infrastructure. And I think what has happened is companies that have done this historically have then ended up with, well, now we have one system that helps us accept the money. We have another system that helps us store the money. We have a different system that helps us disperse the money. Now, when we're trying to create these good user
8:04 Speaker 00 experiences, again, we're stitching these things together. So to use an example of Move, which is the company that has started this large developer conference, if you can provide composable blocks of all the different things that you want to do in payments without any of these middle layers together, engineers are able to build much better products and much more seamless experiences. And then as a result, they are becoming very influential in the buying decision because it's going to drive. The revenue and the product experiences that these companies are able to do.
8:39 Steph Smith? Absolutely. And anytime we see a shift in who the buyer is, that naturally also influences the way products are built. And so maybe you could speak to that. If we now see this buyer, how does that ultimately influence how these products are expected to be built? And maybe also how they're expected to be sold?
8:57 Speaker 00 I used to work with this great VP of engineering and I was on the product side. And what I loved the most is anytime I could come up with the best product experience versus some of my counterparts would be like, well, yeah, this is hard to build. This is not. His expression was always like, well, it's just software. We can build anything. And in financial services and payments, that hasn't always been the case because you've just been limited by how things are stitched together, what systems you can tap into. But if you can imagine, come up with the absolute best product experience. And there's some really unique, different ways to integrate financial services into software platforms. And if you could have the most composable bricks of how to do that, then you can drive the best user experiences.
9:43 Speaker 00 And so I think the product first thinking that's happening in financial services is very much coming to the forefront, both on the consumer side and on the business side, because now we have this great composable architecture.
9:56 Steph Smith? So where do larger financial institutions come into this? Because they're not necessarily known as being the most technically savvy or the most composable code. I mean, I think something that you mentioned in your talk that we referenced at the beginning is that 40% of bank code is still in COBOL, which is pretty legacy. I think you even said that it's not really being taught very widely anymore. And so how do we square that circle of these larger financial institutions? And this walk towards something a little more technologically driven and with developers at the center of that?
10:30 Speaker 00 Yeah. So the most offsite statistic, which I think has stayed very constant, is just the level of banking infrastructure has been around for a long time. COBOL is an old language. There's a lot of code written in COBOL. It is very hard to find COBOL developers right now. There's also a lot of legacy software. Much of financial services is built on enterprise software that has been around since the 60s, 70s and 80s. And the narrative has been that banks are never going to rip a lot of this out. Like it's just way too risky. And I think that is starting to change. I did this talk in 2021 and there was already a ton of momentum towards developers.
11:10 Speaker 00 It's now almost 2024 and there's an entirely new platform shift that's driving even faster change, which is generative AI. And there's a lot of talk in financial services around the risks that are posed by generative AI. There should be more talk around what are the risks of not moving quickly enough towards this trend? And I think we're already seeing that a lot in financial services and in insurance. CEOs from top down are really rethinking their strategy. How can they leverage this platform shift? So how does that impact financial services infrastructure? I think it's challenging this notion that some of this infrastructure will never be replaced because I think what we're seeing is like, take a loan origination system, for instance.
11:54 Speaker 00 There's been new loan origination systems and it's been very hard to get that new tech in. But I think one of the reasons was the new tech was better, but probably not 10x better. Right. Whereas now with Gen AI, it's like, great, we've reimagined the workflow. It's going to take you one tenth as many people because if you take an AML system, for instance, like we're going to automatically write the SAR reports for you. And so the products that are able to come along are so much better that it's challenging the notion that some of this old software is just never going to go. And so I think that's point one. Point two on Cobol, doesn't make sense to entirely rewrite systems in days past, probably not.
12:34 Speaker 00 But if you look at GitHub Copilot, depending on the engineer, that's 30 to 100% increased productivity. So one school of thought is like, well, now you could just rewrite it faster. But you can imagine a world where LLMs get so good that you could feed in large portions of code written in Cobol and they would spit out the same amount of code or the same functionality of said code in a different language. And we're not there yet, but the space is moving so quickly that this change and this shift is very much going to change some of these notions that have been around for a very long time.
13:07 Steph Smith? Totally. It's a very foundational shift. And maybe one other shift I'd love for you to touch on are the regulatory shifts happening at the same time. We did an episode recently where we talked about how different jurisdictions are implementing open banking or sandboxing. And I'm curious to know how you think those also fit into this trend of the increasing involvement of the developer.
13:28 Speaker 00 Well, the open banking trend in particular is pretty interesting. The US is a little bit different than some other countries like the UK and even Brazil that have an open banking mandate from the government that banks have to share their data. And so you can think like, okay, does it make sense for every bank by themselves to independently develop and maintain and do all of these different types of APIs? Or is that an opportunity for broader open source projects, broader standards, such that it becomes a lot easier for financial institutions to maintain some of these open connections and a lot easier for other companies, other software companies to connect into them?
14:09 Steph Smith? Just to round things out, I think it's very exciting that the developer is playing a greater role. But maybe the question that listeners are asking themselves is how does this ultimately impact me? Right? How does the customer ultimately see different kinds of financial services with this developer role becoming more prominent? What do you think this does really to the kinds of products that can be created?
14:33 Speaker 00 I think this is the most exciting part, right? Like you can talk about infrastructure, but really what does better infrastructure do? It enables better consumer and better business experiences.
14:43 Steph Smith? Yeah.
14:43 Speaker 00 I can think of a couple of examples, right? Like there's still 2 billion people that are underserved from a financial services point of view around the world. And there's many reasons for that. But one of the reasons is that if you want to stand up banking infrastructure with all the software, especially in the US, it's going to cost you $10 to $30 a month just to keep the lights on. And that makes sense if you're going to get a mortgage and you're going to get all these different types of products. But it doesn't if you're going to have a low balance account. And so if you now have, and there are several examples, open source cores, where you can just grab the software for free and build around it. You can imagine a very low maintenance cost of financial services, which will open access to many more people.
15:28 Speaker 00 And so that's one theme I'm excited about. And then the second is financial services might have more edge cases than any other industry around, right? Like, the number of different ways that you can pay and different payment types and different sites. Like, it's just, it's huge pain, right? And so if you take, there's now an open source ACH library, and engineers from banks are running trillions of dollars through this library and catching all of these edge cases and fixing it. And so stuff is going to work better, right? If everyone incorporates this hardened library that the best people in the world are testing against, all of us are going to have better products over time.
16:04 Steph Smith? I cannot wait. I think you said in your talk, what the ACH manual or PDF was, what, 500 pages?
16:10 Speaker 00 Yeah, it's about helping sleeping. You can download it.
16:14 Steph Smith? Okay, well, it's great to hear, like you said, that this problem is not only being solved by one company, but that problem is now becoming distributed by the smartest people, the smartest developers. So I think this is a very exciting, big idea. And thanks for sharing with us, Angela.
16:27 Speaker 00 Thanks for having me.
16:30 Steph Smith? Next up, we move to the maritime economy. Technology is tangibly impacting our world on land. Think self-driving cars. In air, think swarming drones. And even in space, think the exponential number of satellites and reusable rockets. But what about the sea?
16:49 Speaker 02 Hey, everyone. My name is Grant Gregory. I'm an investment partner on the American Dynamism team, and this is my big idea for 2024. A new age of maritime exploration. Generations ago, our ancestors took to the seas to explore, and yet today, we know more about the surface of Mars than we do about our own planet's seabed. A new age of maritime exploration is changing that, and founders are leading the way. Maritime faces similar reliability and engineering challenges as air and space. And many of the technologies pioneered by the latest space age can be ready to apply to here, too. Meanwhile, the size and importance of the commercial and defense markets provide substantial reward. We're seeing companies like Flexport, Sironic, and SailDrone, and many others,
17:34 Speaker 02 have already started modernizing maritime. And we anticipate that continued geopolitical supply chain and climate disruptions will further accelerate the demand for change here. Advances in AI, hardware and computer vision present opportunities to transform our cities, ports and trade networks with autonomous, modernized ferries, container ships and fishing fleets. Robots will help sustainably mine precious materials from the seafloor, map and survey waterways, and monitor the health of our ecosystems. New generations of naval and coast guard vessels, ships and submarines will set out to protect our supply chains and our shores. Technology is once again returning to maritime.
18:15 Steph Smith? All right, Grant, you say that we know more about the surface of Mars than we do about our own planet's seabed. This is kind of mind blowing if you really think about it, because we're so close to the sea and it's so integrated into our industries like defense and the supply chain. So, why is this? Why do we know more about Mars than our own seabed?
18:37 Speaker 02 Great question. I think on the surface, it seems kind of silly, but when you think about the raw surface area here, it's quite massive. The ocean covers 70% plus of the Earth. Trade is somewhere north of 55% of global GDP, and over 80% of that by volume is getting transported on the sea. And so you're dealing with something that has, frankly, just substantial scale that we don't have a lot of great line of sight into. If you think about it, when we can look up into the stars, there's plenty of light. If you're looking down into the water, you can only go so far before it fades into black. And so, as a result, historically, we've just had very little literal visibility into the ocean. And most of what we actually know about the ocean seafloor today has been gathered
19:20 Speaker 02 from gravity data from satellites. And so, it's a useful starting point, but the resulting resolution that we have is only about two square miles, I think. And when you compare that to something like our maps of Mars or Venus, those are, by comparison, 50x more detailed. And so, it's pretty astounding, especially when you consider how important the ocean is for life. Ryan Peterson at Flexport has a great line that the oceans are the circulatory system for the world. But fortunately, I think there's a lot of trends that are pointing in the direction of us actually understanding our oceans much more visibly.
19:55 Steph Smith? Totally. And I think you make a great point that it's literal darkness when you get not that deep and our oceans are pretty vast. But maybe you could share a little bit more about the reality of the maritime economy. So when we're talking about the ships or the logistics that we all rely on, What does that look like in 2023? And maybe you can also touch on why it feels like that's been largely untouched by some of the 21st century technology that we're seeing in other industries.
20:22 Speaker 02 I think the one important thing to call out is that maritime is a category that historically has been marked by pretty profound innovation and changes just from a historical perspective. I mean, you started with something like the sailboat, which today is quite de minimis, but at the time was very substantial. You created something like the sextant, the submarine, sonar even, and then something like the shipping container that reduced global costs of transportation for goods by something like 99%. And so maritime in many ways has been the genesis of a lot of pieces of technology that have fundamentally changed the world and give us today. But interestingly, a lot of the current state of the art is very analog.
21:03 Speaker 02 And we're talking about an industry that still operates largely with pen, paper, and maybe emails and phone calls. And I think part of that is just there's an inherent amount of friction in trying to make logistics work and work on time. And as we've globalized, the margin for error has increased substantially, and we're starting to see some additional other pressures in the past few years in a number of ways.
21:30 Steph Smith? Why in 2024 does it seem like there's a true tipping point here to change that equation?
21:36 Speaker 02 The past three years, I think, are a good place to start, which is whether it's COVID, supply chain disruptions, or geopolitical conflict and strife. I mean, all of these things are intimately connected to our oceans and global trade. And so as a result, we've seen some pretty significant dislocations in the past number of years. I mean, during COVID, we had the Ever Given ship blocking the Suez Canal. We had the LA port congestion famously. We've seen a rise in piracy. And then more recently, we had the Titan submarine implosion. And I think they all point to the fact that our supply chains are intimately connected. Maritime is at the heart of that. And the way of conducting things analog is not going to work anymore.
22:20 Speaker 02 And so I think there's that backdrop, which is important to understand. And then there's also some other like larger, I think, deep currents here. One is the International Marine Organization has been establishing new rules and guidance for both emissions, energy usage, and then also broader visibility and data maintenance. And so now ship owners have a real desire and need to figure out how to make their fleet more observable, reliant and performant. And then also you're seeing a real rise on the DOD side with renewed interest in the South Pacific region. And again, that's tied to observability, deterrence, resilience, and a lot of these factors.
23:01 Speaker 02 And then I'd say like on the final part two, encouragingly, we're seeing a wave of startups that are entering more of a maturity stage. Flexport's a great example, and there's plenty of others that have kind of proven out this is a real market and there's real opportunity here to digitize it. And it's difficult, but it's possible. And equally, we're seeing people from a lot of the other American dynamos and companies like SpaceX, Tesla, and Androl. They're taking their learnings and experiences and bringing them over to the ocean.
23:32 Steph Smith? Right. And some of those companies like SpaceX are specifically in aerospace. So what are the innovations, whether it's in aerospace or otherwise, that we can actually borrow now that we're in 2023, heading into 2024? What can be kind of copy and pasted or evolved as we look to maritime?
23:51 Speaker 02 There's a lot. I'd say a great way to delineate it would be there's the processes and understanding how to build quickly and scalably. And then there's the actual technology. And if you're to take an overly simplistic lens to all of this, space and the ocean are both very unforgiving environments. If you have a hole in your rocket, it's going to explode. And if you have a hole in your boat, it's going to sink. And so the status quo here is that there's unrelenting physical forces that are acting adversarily against anything that you're creating and then putting into that environment. And so that creates a dynamic where everything that you're building from an engineering perspective has to be both resilient, hardened, and also needs to be incredibly performant in a whole range of scenarios.
24:36 Speaker 02 And then the water also is just like a very difficult environment, if you think about it. Electronics and water don't mix well at all. There's limited sensing capabilities. A lot of the sensors that we use in other categories like radar and LiDAR, GPS, they don't really work on the water and they definitely don't work underwater. And then also you're just dealing with an environment where everything's moving. You have winds, you have waves, you have currents, tides, boats, and then wildlife. And so all of this needs to be hardened. And I think actually a good example of some of the advances you can see on whether it's machine vision, robotics, autonomy, other capabilities, like Arc Boats has done a great job of encapsulating what the future of maritime looks like in that from a process and team
25:21 Speaker 02 perspective, Mitch and Ryan delivered the Tesla and SpaceX vision and mission to maritime. They're not just creating consumer boats that are electrified. They're actually building better boats that are more user friendly, they're more functional, and they're way more appealing to the average consumer. But while they're doing that, they're also creating a manufacturing process that understands that the factory is the product. And they're doing that in a way that scales beyond just their first Arc One boat, but allows them to eventually expand beyond to other product lines. And I think all that together is quite important. We're seeing plenty of additional, like, interesting developments in underwater imaging,
26:06 Speaker 02 et cetera. It's just space ultimately had to be done, a lot of things in-house and vertically integrated. And we're seeing a similar approach with maritime.
26:14 Steph Smith? Right. I mean, Arc is such a good example. We talked to Mitch semi-recently, and I think I use the term copy and paste, but you brought up a really good point, which is that the aerospace industry got the ball rolling, but we now need entrepreneurs to basically take that technology and specialize it for these different industries. He made a great point that a car will continue rolling down the street for a pretty long while, even if the engine isn't running. A boat will stop immediately, right? So each of these industries is unique. And so maybe you can speak to some of the different applications so folks can get a sense of what may be on the horizon. If we do incorporate these new innovations, what are the kind of geopolitical supply chain or climate applications that
26:57 Steph Smith? you are seeing built or that you see on the way? And maybe also, how far are we in that trajectory? Are these coming in a year, five years, 10 years? What can we expect?
27:08 Speaker 02 There's a lot of change happening and there's a dearth of problems to be solved. I think the good news is that we're seeing a ton of blue space here in terms of what can be done. And in some ways, we're quite early in this chapter, but in other ways, I mean, many of these companies are progressing quite rapidly and have been at it for a while. And so you're seeing some pretty profound changes, maybe on like the observability front. Something that's been really interesting is you have a problem that expands from deep water navigation and autonomy that starts with route optimization, emissions management, but it also ends up in your ports. And so one thing that was pretty mind blowing is that we're dealing with the majority of our trade in the US is tied to the inland
27:53 Speaker 02 navigable waterways. And you have a mix of ocean water and then sediment that's constantly changing the seafloor. And the current way that we navigate all of this is we manually every 100 feet, try and measure this where the bottom of the seafloor is. And as a result, it's changing and it's very inaccurate. And this is something that sonar and other pieces of technology historically have been great at solving. And now we're seeing companies with machine vision and other technologies that are fastening this to the bottom of boats and are doing this in a synchronized manner. And so that you're able to have a live map of where the seafloor is, how many ships can go down a specific port path, et
28:34 Speaker 02 cetera. So that's one. I think the other way to think about this is you're taking a lot of the raw analog instrumentation, you're digitizing that, similar to what Flexport has done with. Customs and freight forwarding and now financing. And then you're also adding sensors, like I was mentioning, to existing vessels and infrastructure. And you're using all that data to capture and create more interesting and efficient route optimizations. And then I think the final thing that's really been interesting is there's companies like Arc, but also Sironic, Anduril with their dive platform that are building new versions of vessels from the ground up with modern day assumptions on how should these vehicles
29:19 Speaker 02 operate in a digital world? And as a result, we're seeing some pretty impressive development on hardware that re-envisioned what something on the water should look like. And so now you don't have to account for people on a boat. You can have something that lasts longer, can go under the water, can focus on mission planning, et cetera.
29:39 Steph Smith? Yeah, and maybe one specific application that you can speak to is just the idea of autonomous fleets, right? So what does that enable? Like you've mentioned in your big idea, the potential for mining or observation. What does the ability, and I know that the idea of autonomous fleets isn't necessarily net new in 2024, but what does that enable? Like, are there second, third order effects of having that autonomy, which we're also seeing increased on land, but instead in water in this case?
30:08 Speaker 02 Yeah, autonomy is interesting because there's so many applications. And to your point, it's been discussed for many years. I think the reality is that there's going to be different waves of autonomy in maritime. There's going to be the actual autonomy for deep water navigation, which frankly is probably less difficult than some of the other problems. Like, once you're in a port, you're navigating against other vessels, against a shifting tide and then also seafloor. And then there's also ideas of like, can we ferry people from one port to another and actually make water transportation and navigation much more efficient and renew that in a way that most people right now don't? And then I think more broadly, on the defense side, or even on like some more of the fringe
30:51 Speaker 02 cases, the goal is to create vessels that can go where people can't. And so whether that's the seafloor and mining precious metals in areas where we can reduce our reliance on other countries for raw imports.
31:07 Speaker 02 Or it's patrolling our coasts covertly in our alleys and protecting them, or just navigating other issues. A lot of this stuff is quite useful. And to date, we've been quite limited based on where people can go. Again, probably the Titan implosion is unfortunately a good example of just the reality that humans can't survive very well under immense pressure in the water.
31:30 Steph Smith? Yeah, that was definitely an unfortunate case. But speaking to that, there are going to be challenges, I imagine, along the way to some of the applications that you're describing. And you already spoke to how going deep underwater is just inherently hard, even when you're on the surface. It's just much more complex than being on land. So what are the key challenges that you expect on this path to modernizing maritime?
31:54 Speaker 02 I think like any industry that has a robust history, it's going to be difficult to change. And there's many things that are kind of tied together. There are structural challenges. There's technological ones. And then there's ultimately challenges with adoption and distribution. When you're dealing with something in the physical world, those are often the hardest things to tackle, and that's where friction is really high. And I think for entrepreneurs, probably first and foremost, the most pressing thing is having a foundational understanding of how the maritime industry works and how your company fits into that picture. Oftentimes, this means finding initial beachhead market and understanding how you can enter
32:37 Speaker 02 that market and then expand into a larger prize. And I think very similar to Amazon, you can't boil the whole ocean at once. You need to thread the needle and go from books to the everything store. And some of the best founders we've talked to in these categories have been able to articulate a very clear understanding of both how the broader ecosystem works and then how their company fits into it. On the technical side, it's a double edged sword because a lot of the tech that we are talking about that can be ported over from other categories like aerospace can be brought over pretty readily, but at the same time, the reality that it is functioning so well is because it is a quite difficult category to build for. And then probably most importantly is, once you build it and you understand what you want to do,
33:21 Speaker 02 you still have to sell this to customers. And this is a category that the people in the industry are quite smart, but it's very difficult to convince them to change because most of the stuff that you're trying to sell to them is dealing with mission critical applications, they can't have downtime, and they're already dealing with huge backlogs and stresses. And so they oftentimes need to have proof of work, they need to be able to see that there's some sort of verifiable evidence that what you're doing will save them money and time. And so as a result, there's just a huge onus on being able to deliver something rapidly in a cost efficient manner.
33:59 Steph Smith? Totally. And by the way, I have to say, I love your water analogies. You said you can't boil the ocean and also vast blue space. So I guess to close things off, there's a lot of opportunity here. That's clear. You see a lot of different companies attacking these kinds of problems. If you were an entrepreneur and there's probably lots listening, I think this is the kind of idea that gets people up in their seats thinking, oh, wow, I can really build something real with my hands. We talk software and hardware coming together. What opportunities are you most excited about? And again, if you were an entrepreneur or builder, how would you even think about starting?
34:33 Speaker 02 I think to start, understand that there's, again, to use these analogies and metaphors, like there's a vast blue space here. And as a result, you need to be really articulate and intentional about what you want to solve. And you're going to be tempted to address plenty of other problems. And the reality is having focus here is going to be paramount to a successful company. And then from there, you're going to earn the ability to operate and expand and do much more. I think another interesting idea that we've seen is that in categories that are really resistant to adopting software, oftentimes hardware is a great Trojan horse where you can sell that the customer base is used to purchasing some form of hardware. And then from there, you're able to layer in software capabilities
35:18 Speaker 02 and use that as the wedge from which you can kind of expand. But I think to carry on the autonomous vessel and vehicle side of things, we're seeing in defense and in many categories, there's an asymmetric advantage to having high volume and low cost. And so in this world of defense, but also just observability, I anticipate there's going to be a number of companies that are able to provide swarms of vessels, both on the surface and also subsurface, that are able to not only protect our shores and our allies, but are able to detect environmental damage, issues in our oceans, and ultimately able to help us probably observe our fisheries and other supply chains much more effectively.
36:01 Steph Smith? Absolutely. Well, I hope we have this conversation again soon. And that quote that we discussed at the very top will have flipped, where we at some point know more about our own ocean seabed than we do about Mars. Or hopefully both. We know a lot more about both next time. But Grant, thank you so much. This was great.
36:20 Speaker 02 Thanks so much.
36:23 Steph Smith? Finally, as software continues to eat the world, what do we do with all that data generated? Specifically, can unlocks in computer vision make sense of the vast amount of video data we collect, potentially modernizing industries from transportation to agriculture and even to mining?
36:41 Speaker 01 Hi, I'm Kimberly Tan. I'm an investing partner at Andreessen Horowitz on the American Dynamism and the Enterprise team. And this is my big idea for 2024. I believe that in 2024, we'll likely see new applications of computer vision and video intelligence in the real world. I think leveraging insights from video data has become pretty commonplace in the enterprise to help companies make better informed business decisions. But this capability could be even more powerful in the real world, given how the rich and comprehensive the nature of the available data is. However, many industries today still lack modern systems to actually capture and make sense of that video. And oftentimes, customers actually have no existing video infrastructure
37:23 Speaker 01 or use pretty legacy video systems that are difficult to integrate if you actually want to use modern software. So we see businesses tackling this problem by leveraging a hardware and software model, allowing them to sell both the hardware video cameras as well as the software to customers. These businesses often tailor their go-to-market approach to target a very specific customer, investor of those particular needs. So for instance, companies like Flock Safety, and for just examples of this hardware and software model, where they've built large businesses around keeping residential neighborhoods and schools safe, respectively, for example. And we think that the same success could be found in other industries, such as, for example, transportation, industrials, agriculture, mining.
38:06 Steph Smith? All right, Kimberly, so this, in a way, is a whole new twist on the idea of software is eating the world. In this case, it's software is eating the physical world, in a way. So maybe you could just give us a sense of just how much video is actually currently being captured?
38:21 Speaker 01 Yeah, definitely. And by the way, I think software eating the physical world is a great articulation of the broader American dynamism thesis that we have at the firm that we're really excited to invest behind. I would say there's a vast amount of video data that is being captured, just given the general prevalence of cameras, either on smartphones or just in a lot of places that we live our day-to-day lives. And I think the real question is how much of that data is actually being captured and utilized in some interesting way versus just passively existing on different devices and not actually being analyzed or processed in any way.
38:53 Steph Smith? Yeah, I think that makes sense. And I think we're all very familiar with how much data each of us individually is capturing. But what unlocks in the hardware or the software to process the data actually allow us to better leverage this video and also maybe have contributed to just how much data is being collected?
39:10 Speaker 01 I would say like, the reason why I'm particularly excited about this thesis in the next year or so is I feel like it's the confluence of a lot of things that are finally good enough, that were always on the cusp of being possible, but now have hit some cliff that we can see really large venture-scale outcomes. For example, the proliferation of cameras and availability of videos. There's just more cameras getting installed for different use cases where the insights aren't being leveraged as widely they could. So like I mentioned, either on people's smartphones, there's all these home Nest cameras that are getting built out, there's cameras in people's workplaces, traffic cameras, et cetera. And I think the availability of actually just even having the data there is one thing that we're reaching a tipping point on.
39:53 Speaker 01 The second is, I would say, that tech just becoming broadly more... Cost efficient and available. So first through like cloud cost dropping such that you can actually do more computation in the cloud, which has historically been a barrier, given just how much data there is residing in video and how expensive it is to process that data. And then secondly, on that point, increases in edge computing capacity through cheaper chips and camera quality per dollar, allowing you to run more complex and larger models on better cameras at lower cost and thus reducing any sort of bandwidth constraints you would have had in actually sending it to the cloud. So I think the tech becoming more efficient is a second reason why we're really excited.
40:38 Speaker 01 The third would be just broadly innovation on the models themselves. So everybody I feel like now knows transformers and how exciting it's been for like generative AI innovations that we're seeing in things like ChatGPT. And while transformers were designed for language tasks initially, they've actually been applied in vision applications as well and works quite well for things like image classification and object detection versus the status quo previously. So I think seeing transformers applied in the vision space will be really exciting. And then, once you have the video, being able to query it and actually like look at what is in that video and being able to just see through using something like generative AI, where you can query a natural language instead of having to write a bunch of code to actually look through it and understand it in a meaningful way.
41:24 Speaker 01 I also think there's just like a business model understanding question, which a lot of people thought for a long time that selling into these difficult verticals was tough. And I think now that we have companies like Flock Safety proving that it is possible, people can apply those lessons into other verticals where it hasn't currently been applied yet.
41:40 Steph Smith? Absolutely. So it sounds like there is an element of the cost curve changing, elements of innovation and specifically LLMs changing the game and maybe through that, business models evolving. Could you speak to maybe some of the applications that you think can emerge here? I mean, you mentioned transportation, agriculture, industrials and mining. Give us a sense of the types of applications that might sprout from these different factors coming together.
42:06 Speaker 01 I want to preface by saying there's probably like way more than we can even envision. So these are just examples that I think are exciting. But in transportation, for example, there's a lot of back and forth that happens with just essentially confirming that you're giving the right item to the right person at the right time and charge the correct amount. And today, a lot of that is done through very manual emails, invoice processing, reconciliation, confirmation of receipt of items, compliance on specific tariffs, et cetera. And a lot of that through computer vision, you can identify that this was the item that we gave to this person at this time and just cut out a lot of that inefficiency. I think industrials and other pretty large scale industries like that, there's a lot
42:47 Speaker 01 of labor or OSHA compliance on workplace safety and conditions that have to be met that inspectors come by, you often have to fill out all these checklists and make sure. And through computer vision, none of which requires actually monitoring the people themselves, you can make sure that the conditions are safe for the folks who work there. And then in something like agriculture, the bread and butter of that industry is livestock or crops or whatever you're growing. And I think there's a lot of insights that you can be gained without having to have somebody manually looking at it all the time about making sure that the weather conditions are correct or the crops are healthy and such.
43:20 Steph Smith? Yeah, I think that makes a lot of sense. And I think people can easily imagine those use cases. But what about the use cases where this video is deeply ingrained in our everyday lives? I think the natural question as it relates to that scenario is how we avoid the kind of intelligence that is a slippery slope into a big brother scenario or something like that. How would you calm those concerns? Or how do you see regulation potentially playing a role in this video centric world that you're describing?
43:49 Speaker 01 Yeah, I think this is a really important question and topic that all companies in this space and we monitor pretty closely. I think privacy will continue to be a very important question. It's hard to say how regulations will shake out, but there's a lot of things that companies can do around both data rights and data privacy concerns to make sure who actually owns the data rest. There's lots of things around how long you're actually allowed to store the video for. And even in computer vision itself, like what are you tracking? Are you tracking, for example, just the license plate? Are you tracking just the position of a certain piece of equipment at a certain time instead of tracking any individual person? So I think there's a lot that can be done to make sure that
44:32 Speaker 01 people's privacy rights are protected while allowing us to use the technology in a way that makes people's lives better. But it is one where I think companies have to work very closely with the stakeholders on both the regulatory as well as the personal and business side to make sure that everyone feels comfortable.
44:49 Steph Smith? Absolutely. And of course, this is a big idea for 2024. So we spoke a little bit to this and the different trends coming together. But why do you think this is the year where we're really going to see a tipping point in terms of widespread adoption? And maybe you could just speak a little bit more to what you hope to see builders come and work on this coming year.
45:08 Speaker 01 We've talked a little bit about this before, but I think it's just a lot of things coming together to make this the right time to do it, both on the cost side. I think there's a little bit of like a labor element of labor shortages in a lot of the industries that we were talking about that maybe opened the door to software innovation where people were not as used to buying software in the past. I think that the general focus that people have on AI today naturally lends itself to the question of where else can AI be applied? That isn't what we've been seeing with all the exciting generative creative stuff and thinking about how these technologies can actually work in these more brick and mortar, physical use cases. And then I think, yeah, like a lot of companies have now grown up to be successful companies that have employed this model.
45:51 Speaker 01 And a lot of new founders who are starting companies looking to see which companies they admire will have these companies as models to potentially emulate or take lessons from, whereas that wasn't true a couple of years ago.
46:03 Steph Smith? Yeah, absolutely. And we spoke about this at the very beginning, but it does feel like software is penetrating more and more of our world since the decade that has passed, since Mark originally wrote that piece. But this idea that software is definitely penetrating the physical world is ringing very true and feels a little unstoppable. So it's going to be very exciting to see how different builders leverage this new technology and focus on video.
46:26 Speaker 01 Definitely. We're very excited to see it as well.
46:29 Steph Smith? All right. I hope you enjoyed this full three-part series. If you did miss part one or part two, make sure to go check those out as we explore seven more big ideas in depth. Plus, if you want to see our full list of 40-plus big ideas for 2024, you can head on over to a16z.com/bigideas2024. By the way, there you'll also find all our crypto big ideas, which we didn't cover in audio. But if you do want to learn more about them, you can also go check out our sister podcast, Web3 with a16z, which also covered these crypto big ideas in their previous episodes. I really hope these big ideas get you excited about just how much change is in the air and also how much opportunity there is still to build. And if you're listening
47:11 Steph Smith? to this now, why wait till 2024? It is time to build.
Previous Tony Robbins on the Future of Health & Longevity Next Big Ideas 2024: AI Interpretability: From Black Box to Clear Box with Anjney Midha Page updated September 30, 2026 · transcript and summary are machine-generated
