Main Topic: The podcast episode is a recap of the past year, highlighting the advancements in various industries, including AI, autonomous vehicles, electric boats, and fintech. The episode also explores the opportunities and challenges in these industries and how they are shaping the future.
Key Points:
Participants:
Takeaways:
As we close off 2023, we wanted to revisit some of our favorite episodes from the last year. From AI hardware, to healthcare regulation, to disrupting the world’s largest asset class, to the electrification of nearly everything… there’s something in here for all listeners.
Resources:
AI Hardware, Explained:
https://a16z.simplecast.com/episodes/ai-hardware-explained
Chasing Silicon: The Race for GPUs: https://a16z.simplecast.com/episodes/chasing-silicon-the-race-for-gpus
The True Cost of Compute: https://a16z.simplecast.com/episodes/the-cost-of-compute
AI Revolution: Disruption, Alignment, and Opportunity: https://a16z.simplecast.com/episodes/ai-revolution-disruption-alignment-and-opportunity
The Robot Lawyer Resistance; https://a16z.simplecast.com/episodes/the-robot-lawyer-resistance
The Quest for AGI: Q*, Self-Play, and Synthetic Data: https://a16z.simplecast.com/episodes/the-quest-for-agi-q-self-play-and-synthetic-data
The Road to Autonomous Vehicles: Are We There Yet?: https://a16z.simplecast.com/episodes/the-road-to-autonomous-vehicles-are-we-there-yet-PstZz2G_
The Electrification of Everything: From Sky to Sea: https://a16z.simplecast.com/episodes/the-electrification-of-everything-from-sky-to-sea
When Two Giants Intersect: Healthcare Meets Fintech: https://a16z.simplecast.com/episodes/when-two-giants-intersect-healthcare-meets-fintech
Salary Transparency: Clarity or Chaos?: https://a16z.simplecast.com/episodes/salary-transparency-clarity-or-chaos
Disrupting the World’s Largest Asset Class with Adam Neumann: https://a16z.simplecast.com/episodes/disrupting-the-worlds-largest-asset-class
The Data Highway Above with Privateer’s Steve Wozniak, Alex Fielding, and Dr. Moriba Jah: https://a16z.simplecast.com/episodes/new-the-data-highway-above-with-privateers-steve-wozniak-alex-fielding-and-dr-moriba-jah
Stay Updated:
Find a16z on Twitter: https://twitter.com/a16z
Find a16z on LinkedIn: https://www.linkedin.com/company/a16z
Subscribe on your favorite podcast app: https://a16z.simplecast.com/
Follow our host: https://twitter.com/stephsmithio
Please note that 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. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures.
Stay Updated:
Find a16z on X
Find a16z on LinkedIn
Listen to the a16z Podcast on Spotify
Listen to the a16z Podcast on Apple Podcasts
Follow our host: https://twitter.com/eriktorenberg
Please note that 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. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures.
Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.
Steph
The speed at which we get to an AGI is dramatically faster.
Guest 1
Your boat is charging your house.
Guest 2
This is a great example of an act that actually incentivized a whole bunch of startups to come out of the woodwork.
Guest 3
This requires a deep, deep understanding and tremendous amount of machine learning.
Guest 4
Your biggest asset is your current employee population.
Guest 5
The engineer says, feet on the ground? What can I actually do and build and deliver to people?
Guest 6
Hugely controversial. There's still lawsuits in play.
Guest 7
You could look across all these industries and find those opportunities where the technology that exists today delivers a better and more compelling experience to people.
Guest 8
Whenever I tried to see the future a year ahead, I knew it one year ahead because I was working on it.
Guest 9
We have about a five-year window, I think, as a society to kind of figure out what this means and how to adapt.
Steph
Hi, everyone. This is Steph. Welcome back to the A16Z Podcast. I cannot believe we are at the end of 2023, and I also can't believe it was about a year ago today that the A16Z podcast officially relaunched. And since then, a lot has changed. I mean, generative AI tools have absolutely exploded in popularity, fully autonomous vehicles now roam the streets of San Francisco, and entire industries from the skies to the seas are being electrified as we speak. And those are just a few of the groundbreaking innovations that we did cover here on the podcast.
We were also joined by some incredible guests, including a Nobel laureate who is now the lead scientist on the James Webb Space Telescope. We also had founders hailing from the companies building some of the most important large language models today. We had founders trying to give billions of people internet access. The original creators behind one of the world's largest digital influencers. And even in our latest episode, Mark and Ben got to sit down with the one and only Tony Robbins.
So as we round out this year, we wanted to recap some of our favorite conversations, highlighting how our collective future is being defined by the builders advancing technology day by day. Today, you'll hear from the likes of Mark Andreessen and Steve Wozniak, and if any of these episodes do pique your interest, we'll include the links to the full episodes in our show notes. So thank you so much for listening this year. And without further ado, let's dive in.
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 slash disclosures.
2023 was a year dominated by AI. From healthcare to code, AI is in the midst of disrupting some of the world's largest industries, forcing even multi-billion dollar companies to rethink their strategy. But as AI software became more important than ever, the hardware running it has followed suit. Here in our first clip from our AI hardware series, you'll hear from Guido Eppenceller, A16Z's very own special advisor on AI infrastructure, who also happened to be the former CTO of Intel's data center and AI unit. Here, Guido breaks down the GPU architecture and why it's so game-changing for AI processing, plus why NVIDIA has been leading the pack here so far.
Guido Eppenceller
If you're running any kind of AI algorithm, this AI algorithm runs on a chip. And the most commonly used chips today are AI accelerators, which are, in terms of how they're built, actually very close to graphics chips. So the cards that these chips are on that are in these servers are often referred to as GPUs, which stands for Graphics Processing Unit, which is kind of funny, right? They're not doing graphics, obviously, but it's a very similar type of technology.
If you look inside of them, they basically are very good at processing a very large number of math operations per cycle in a very short period of time. Very classically, an old-fashioned CPU would run one instruction every cycle, and then they had multiple cores. So maybe a modern CPU can do a couple of 10 instructions. But these modern AI cards can do more than 100,000 instructions per cycle. So they're extremely performant.
So this is a GPU, and these GPUs run inside of servers. Think of them as big boxes. I have a power plug on the outside and a networking plug. And then these servers sit in data centers where you have racks and racks of them that do the actual compute.
Steph
Let's quickly recap. CPU is central processing unit and GPU is graphics processing unit. And while both CPUs and GPUs today can perform parallel processing, the degree of parallelization is what sets GPUs apart for certain workloads. So for example, CPUs can actually do tens or even thousands of floating point operations per cycle, but a GPU can now do over 100,000.
Guido Eppenceller
The basic idea of a GPU is that instead of just working with individual values, it works with vectors or even matrices, right? Or tensors, more generally. TPU, for example, is Google's name for these kind of chips, right? And they call them tensor processing units, which is actually a pretty good name for them. The cores in these modern GPUs are often called tensor cores, because they operate on tensors. And basically the core of their value proposition is they can do matrix multiplication. So remember matrix like the rows and columns of numbers, they can, for example, multiply two matrices in a single cycle. So they're very, very fast operationally. And that's really what gives us the speed that's necessary to run these incredibly large language and image models that make generative AI today.
Steph
Today's GPUs are far more powerful than their ancestors. Whether we're comparing to the earliest graphics cards in arcade gaming days 50 years ago, or the GeForce 256, the first personal computer GPU unveiled by NVIDIA in 1999. But is it surprising that we're seeing this chip design applied so readily to the emerging space of AI? Or should we expect a new architecture to evolve and become more performant in the future?
Guido Eppenceller
In one way, I think it's very surprising. Who would have thought that my gaming PC or my Bitcoin miner would eventually become a good AI engine? But at the same time, what all of these problems have in common is that you want to execute many operations in parallel. So you can think of a GPU as something that was built for graphics, but you can think of them also just as something that's very good at performing the same operation on a very large number of parallel inputs. A very large vector or a very large matrix.
Steph
All right, so perhaps it's not so surprising that NVIDIA's prized GPUs are aligned to this AI wave. But they're also not the only company participating. Here is Guido breaking down the hardware ecosystem.
Guido Eppenceller
The ecosystem comes in many layers. Let's start with the chips at the bottom. NVIDIA is king of the hill at the moment. A100 is the workhorse that powers the current AI revolution. They're coming up with a new one called the H100, which is the next generation. There are a couple of other vendors in the space. Intel has something called Gaudi, Gaudi 2, as well as their graphics card with Arc. They're seeing some usage. AMD has a chip in this space. And then we have the large clouds that are starting to build, or in some cases have been building for some time, their own chips. Google with the TPU you mentioned before, it is quite popular. And Amazon has a chip called Trainium for training and Inferentia for inference. And we'll probably see more of those in the future from some of these vendors. But at the moment, NVIDIA still has a very, very strong position as the vast majority of training is going on on their chips.
Steph
And when we think about the different chips, you mentioned like the A100s are the strongest, and maybe there's the most demand for those. But how do they compare to some of these chips created by other companies? Is it like double the performance or is there some other metric or factor that makes them much more performant?
Guido Eppenceller
It's a great question. If you look at the pure hardware statistics, so how many floating point operations per second can these chips do? There are others that are very competitive with what NVIDIA has. NVIDIA's big advantage is that they have a very mature software ecosystem. Imagine you are an artificial intelligence developer or engineer or researcher. You're often using a model that's open source that somebody else developed. And how fast that model runs, in many cases, depends on how well it's optimized for a particular chip. And so the big advantage that NVIDIA has today is that their software ecosystem is just so much more mature. I can grab a model, it has all the necessary optimizations for NVIDIA to run out of the box. I don't have to do anything. But with some of these other chips, I may have to do a lot more of these optimizations myself. And that's what gives them the strategic advantage at the moment.
Steph
We just heard about the underlying infrastructure powering AI's exponential growth. But with this growth comes a lot of questions, like how will AI change the way companies operate? Will it automate creative tasks? And does it need to be regulated? Our next rewind is with Dylan Field, co-founder and CEO of Figma, a browser-based web design tool tentatively acquired by Adobe for $20 billion. Dylan was one of the many speakers at our AI Revolution event, alongside founders of OpenAI, Anthropic, Character.AI, Roblox, and more. Now, as someone who helped reshape the entire design landscape over the past decade, here Dylan weighs in on how AI may impact the design process and also answers the $1 billion question. Will AI eventually replace us?
Dylan Field
It's kind of interesting. It puts in the question, is there going to be less things to design, or is AI going to do all the design work? You're on one of those paths. On the first one, will there be less things to design? If you look at every technological shift or platform shift so far, it's resulted in more things to design. You got the printing press, and then you have to figure out what you put on a page. And more recently, mobile. You would think, okay, less pixels, less designers, right? But no, that's when we saw the biggest explosion of designers.
If you'd asked me this at the beginning of the year, you might have said, okay, we'll all have these chat boxes and people will be asking questions in them and that's going to be our interface for everything. Look at OpenAI. They're on a hiring acquisition spree, trying to get product people and designers right now so that they're able to make great consumer products. It turns out design kind of matters.
The second one, will AI be doing the design? I think that's pretty interesting. So far, we're not there. Right now, we're at a place where AI might be doing the first draft. Getting from first draft to final product turns out that's kind of hard. And usually it takes a team. But if you could get AI to start to suggest interface elements to people and do that in a way that actually makes sense, I think that could unlock a whole new era of design in terms of creating contextual designs, designs that are responsive to what the user's intent is at that moment. I think that'd be a fascinating era for all designers to be working in, but I don't think it replaces the need for human designers.
Steph
Design isn't the only industry being redefined by AI. In 2023, DoNotPay, an online platform and chatbot for legal help, attempted to bring AI lawyers into the courtroom, sparking a conversation about the role of AI in law. Our next clip is with the founder of DoNotPay, Joshua Browder. Joshua has spent years building products that automate the role of lawyers, helping consumers fight legal battles and dispute unfair charges, from parking tickets to excessive medical fees. Now, with the growing accuracy of AI chatbots, the entire legal landscape looks pretty ripe for innovation. And Joshua is helping to pave the way. Here's his vision for the future.
Host
I want to ask maybe a far-fetched question, but if we look super far ahead, how do you see this new technology, AI in particular, reshaping the legal landscape? As an example of this, when might a robot lawyer actually write law that governs the implementation of AI within law? That seems like it hasn't happened yet, but something that maybe is inevitable.
Joshua Browder
I think in the next year, it's already secretly writing laws. A lot of laws originate from an overworked staffer in Congress. I'm sure it's 2 a.m., they need to finish it, and they're probably typing in ChatGPT, asking it to help them a bit. So I'm sure in the next year that's going to happen. In the future, I think a lot more court systems will be automated and tools like DoNotPay will help people. So I think it's a very exciting time.
Host
Speaking of an exciting time, I feel like DoNotPay has done an excellent job of keeping up with the times. Something I saw you build recently is the ChatGPT extension, which you can basically understand the terms and conditions of a website more effectively. So how are you thinking about the gaps that still exist for consumers and maybe what other opportunities there are to give consumers their rights back?
Joshua Browder
When new technology comes out, we think, how can it help people? Terms and conditions is a great example because people can't understand the changes being made, what they're giving up. This is actually our first public product relating to true AI, where you can paste in any terms and conditions, upload a document, including leases, and it will go through and tell you all the things that are non-standard and also things you should watch out for. That's really important because people don't have time to read a 100-page document, which is what some of these leases are, especially in New York. So that can be really helpful. We're trying to be like David versus Goliath so that AI can turn David into Goliath versus Goliath and help ordinary people.
Host
Something that's coming up in my mind as you're speaking is a question around why the law is so complex. Maybe this is such a silly question because there's nuance to life and therefore there's nuance to law. But why are so many laws 100 pages long that no regular consumer can actually read through and understand? Should it not just be a rule system that you could even effectuate through code, for example?
Joshua Browder
I think there's a lot of lobbying that goes on. The big companies spend huge amounts of money shutting consumers out of their rights. Every year it seems to get worse. There are lots of positive laws that have come out, like the CCPA and other laws around that. But I think big companies have a lot of influence in creating the laws. One area that I'm worried about is anti-AI laws. One could imagine a law that says that no bots are allowed unless they say they're a bot. You're already seeing companies implement these policies, and that could easily be a law. I could imagine members of Congress signing up for that idea. And that would hurt a lot of consumer rights and automated systems helping people.
Host
What makes you think that people will actually implement a law like that? Or are there other rules that you see on the horizon based on your experiences with companies with lawyers who are pushing back against some of this new technology?
Joshua Browder
The fear when any technology comes out is huge. And it doesn't help that people like me are stirring things up. Fear creates action on behalf of governments and big companies. So that's very worrying. The good news is that there are a lot of things in the law. We have over 100 years of law to deal with, that if you jump through the right hoop and follow the correct processes, you can get results. One thing we automate is credit report disputes. They say that if you submit a correct letter and it's signed in the correct way and all of this stuff, you can dispute something on a credit report. This allows us to get old or inaccurate items off people's reports. Even if it's coming from AI and we have to tell the big company that this was written by AI, it's not going to be influenced. They still have to respond to that letter. There are all these rights that people have, and I don't think that they can strip them all away with one law, but it is worrying about where Congress is heading.
Steph
As the industry continues scaling and refining these models, there are still many open questions on the horizon. In a recent episode with A16Z general partner Anjaneh Mittal, we explored some of the most important frontier research areas, from synthetic data to self-play. But first, here Anjaneh clarifies why math problems, even at the grade school level, are a suitable testing ground for this kind of research.
Anjaneh Mittal
The reason grade school math is so interesting is because it's pretty well scoped and you can ask the model to break down a word problem. One of these classic, "Jimmy has 10 apples and John has 16 apples. If they combine their apples, how many do they have in total?" You can ask a model to break down its reasoning step by step. When it does that, it doesn't just give you the end answer, it gives you all the series of steps in the middle that it took to get to the final answer.
What you can do with reinforcement learning applied to chain of thought prompting is you can say, instead of traditional reinforcement learning from human feedback, where you just asked a human grader to grade the outcome of the model, the final step, or you ask another AI system to grade the final step and say, "Is this answer right or wrong?" You can actually have an AI now score every intermediary step and score whether each step of the way was correct or not. Then what you get is a much more granular set of rewards.
In the case of cake baking, instead of just scoring whether the end cake was good or not, you can start scoring the individual steps. Did they bake it at the right temperature? Did they put the right amount of dough in? The big question is, does that allow the model to start learning about reasoning, not just producing some end outcome? If you can get it to reason, then it can start solving, planning, and doing higher order thinking of a kind that humans have.
These techniques all play together. Reinforcement learning traditionally has been a way to get these models to improve. Cue learning or model-free learning is a way to generalize reinforcement learning beyond just well-enumerated systems like games. And lastly, synthetic data and self-play come into relevance here because when you're having an AI score each individual step of your reasoning, you're generating a bunch of really valuable data that you can then train the system on that's synthetic. It's not generated by a human, it's generated by an existing system. These are the big components that are working together to potentially ultimately produce a model that's 10,000 times or 100,000 times better than GPT-4.
Host
Let's double click on that element of self-play and potentially synthetic data. To underscore that, right now many of these models do have an element of reinforcement learning, but that comes from human feedback. RLHF is one of the methods that these different labs are using to orient the existing models, but obviously that requires humans in the loop. What you're getting at is potentially the ability to obfuscate that and have AIs in the loop. Can you speak a little bit more to that idea of self-play and the ability to create synthetic data? What does that unlock other than just scale and speed? Or is that enough for us to get to those next levels?
Anjaneh Mittal
Scale and speed are pretty valuable. If you go back to first principles on how these models are trained and get better, the rate at which they improve is remarkably predictable. They're predicted by a set of AI scaling laws that basically say you've got three ingredients to producing machine intelligence: compute, data, and algorithmic innovation. I think what self-play and synthetic data do is they allow the scaling of these models much more rapidly. Whereas with reinforcement learning from human feedback, you're constrained by the number of humans you can have provide that feedback. Synthetic data, and in particular the types of AI feedback scoring that we're talking about here, is orders of magnitude more scale. It's not 2x or 3x. The sheer constraint is actually how much compute do you have to run these AI models to do all this scoring.
To make this a little bit more concrete, it might be helpful to talk a little bit about reinforcement learning from human feedback.
Host
Please.
Anjaneh Mittal
The 1.0 version of this world. A common example of reinforcement learning from human feedback is a system like ChatGPT or DALL-E or Midjourney. When Midjourney generates an image or when ChatGPT generates a message for you, there's a little thumbs up, thumbs down button. They track when users give you a thumbs up or thumbs down. They use the number of times people give thumbs up or thumbs down and use that as a signal for reinforcement learning to improve the kinds of messages it gives you next time around.
The reinforcement learning from AI feedback, in this case, replaces that thumbs up and thumbs down from users like you and me, and instead substitutes that with increasingly smarter AI models that these labs have built to then provide much more granular scoring than just the thumbs up, thumbs down. You can now start providing a score of 1 to 10 on the intermediary steps. You're right, scale is really the biggest unlock here. When you replace humans with models that provide this feedback 24-7 at orders of magnitude larger volume than we could do with just humans, then the speed at which we get to an AGI is dramatically faster. So any breakthrough in those systems represents, I think, a nonlinear increase or acceleration towards this future.
Steph
And while many people think of chatbots or personalized avatars when they think of AI, the world of applied machine learning is also catching steam. One way: self-driving cars. In 2023, Waymo vehicles covered over 700,000 trips. Yes, you heard that right, 700,000 trips and millions of driverless miles. They also received their final permit from the California Public Utilities Commission, signifying a huge step forward. Up next is a clip from our episode with Waymo's chief product officer, Saswat Panagrahi. As CPO, Saswat has been at the forefront of this industry's unique combination of technical, regulatory, and economic challenges, from ensuring rider safety to building defensible moats through hardware. Now, in this clip, recorded in April, Saswat discusses where we really are in the shift to autonomous vehicles, and the challenges that still lie ahead.
Host
It's felt like this is coming for a while. I almost feel like the future is here. We're seeing this, we're sitting in a car with no driver. Maybe you could map out where we are in that arc, the five levels of autonomy, and also where we still have left to go.
Saswat Panagrahi
Certainly. We are in level four right now. Fully autonomous. As you can see, nobody in the front seat, no expectation of a human to take over. In level two and three, it's really crucial to communicate the expectations to the driver. Because it's very easy that during a normal situation, the driver feels, "Hmm, the car is driving well, so I can pick up a book and start reading." No. That's serious. There is an expectation to take over. So we are in that level four with a certain scope. Right now, if it were to begin heavily snowing in SF, which it has, believe it or not last season there was a little bit of snow, we wouldn't operate there. We couldn't. But level five is truly defined as anywhere, anytime.
Host
The idea of autonomy, people kind of view in this binary way. The whole idea of level puts that into perspective. Level two is lane sensing, automatic braking. Is level three basically what we're doing now, but there would be a human in the front? Or how would you define level three?
Saswat Panagrahi
No, no. I would say vast difference even between level three and level four. Huge. It's almost like the difference between driving and flying, I would say.
Host
Okay.
Saswat Panagrahi
It's a massive difference. Because this concept still that within seconds you need to take over versus now we can have this conversation. We're not doing anything. That would be challenging, right? The assurance level you need to get to for level four is just a universe different, I would say.
Host
Tell me a little bit more about the barriers along the way to level four. What would you say? Was it the technology, the regulations, some combination? Are there major factors that have really delayed us from getting to this point where we're now at level four?
Saswat Panagrahi
On any journey, when you're trying to do something that has never been done before, there will always be ups and downs. Some challenges you foresaw, some you didn't. The key is, are you clear on how massive the prize or the benefit to society is on the other side of it? Because then that makes it all worth it. We were super clear from the get-go that a fully autonomous car that does not get drowsy, that we even have physical constraints. Even when you are alert, let's say you were looking for parking on this side of the street, your face would be turning towards that, constantly looking for that parking spot. So you wouldn't be able to see...
Host
You don't have your sensor.
Saswat Panagrahi
Exactly. You can't keep turning back and forth. So we're fundamentally convinced that a fully autonomous driver is going to be safer. Once we started that, the largest challenge I would say was largely technology, but I mean two different things. One is building the driver itself that can drive under these conditions and having the high grade of performance. But also measuring that is pretty hard.
This smooth early stage drive under very tight constraints is now, with today's technology, not terribly hard to build. But to be able to do that at the scale of 24-7, busy intersections at slow speed, but also high-speed intersections of Phoenix. In Phoenix, for example, the streets are wider, so you don't deal with these narrow situations. But the driving speed is 45. People are sometimes going 60 on that. That means you've got to see a lot further. Very different sets of challenges, very diverse ones.
The technology really required the full stack. We build the hardware, we build the software. Because if you built just the software and waited for somebody else to deliver the hardware, the speed of learning, the speed of iteration that was necessary to build something like this was so steep that it was not feasible. So we had to build the lasers, the cameras, the radars, and the software on top, and the massive simulation infrastructure as well.
Host
I was going to say, there's so many moving parts. Maybe let's talk about that technology. I see a bunch of different appendages to the car. Maybe you could break that down. What is happening? How is all this technology coming together? And what are the bits and pieces that you've added onto the car that allow it to be autonomous?
Saswat Panagrahi
Fundamentally, you can think of it like, is the car aware of what's happening around it? And then can it anticipate what the things around it are going to do?
Yeah. And then reasoning on what it should do. These are sort of the three components. In perceiving what's around us, think of the example we were just discussing. You were trying to look for a parking. So you're focused on that task. This car with those appendages, as you mentioned, can see three football fields away, 360 degree, and it's getting a snapshot multiple times a second.
It's relying on a combination of the state-of-the-art laser, camera, radars, all strategically positioned. To give you an idea, lasers give you a very precise understanding of everything around you. It's the smallest detail. So if there was a child an inch out of this pole, it would be able to mark, oh, this is a demarcated child away from that pole. It would get to see that. But the cameras are needed to distinguish between the red light and the green light.
Host
Oh, yeah.
Saswat Panagrahi
And the radars can almost see around corners, even when the laser and the camera or our human eyes can't. Because they can sense objects coming in. We took an approach that we want to combine the best strengths of each of these modalities to create the best picture of what you can see around the world. You're just incredibly better than a human possibly could, both due to their attention span, the range, the fidelity, and the combination of these sensors coming together. So that's what we see.
But then there's a harder challenge of anticipating what the person will do. Take a look at that pedestrian. They're standing pretty close to the crosswalk, right? But you don't know if they're going to move. Are they going to jump in or are they going to stay? Are they going to jaywalk or are they going to obey the light? This requires a deep, deep understanding and tremendous amount of machine learning.
Host
Cars are not the only mode of transportation that have gone through a pivotal moment over the last year. Biden's Inflation Reduction Act was the single largest climate investment in U.S. history. And its second-order effects are already becoming increasingly evident. Everything around us, from the planes in our skies to the boats in our seas, is in the process of becoming electrified. Our next clip is with Mitch Lee, co-founder of Arc Boats, a company manufacturing electric boats for water sports to start, and Gregory Davis, CEO of Aviation, a company building planes with an all-electric design. Together, they discuss the opportunities in our all-electric future.
Guest 1
There's plenty of these product opportunities you can look across all these industries and find those opportunities where the technology that exists today, the electric powertrain, delivers a better and more compelling experience to people. I think the three of us are all doing that. There's also supply chain opportunities. We want our battery prices to be lower. We want our high voltage electrical power chain components to cost us less. There's a tremendous number of opportunities at that kind of base supply chain level.
And then there's this third category of what I call second order opportunities. You're going to have an electric future. Imagine what that looks like. Are you going to be towing around a bunch of electric boats in electric trucks? Or are you going to be towing around an RV in an electric truck? What does that mean in terms of the new charging infrastructure you need to go support that? Or the new technology you need to go make that happen? Maybe it's powered trailers that can actually help the vehicle. There's just a ton of opportunities. If you cast your mind forward a few years and think about what are the new problems that are going to exist when the marine industry is going electric and the aviation industry is going electric and the bus industry is going electric.
Guest 2
Yeah, and I think interoperability and standards, industry standards, is a challenge today. We've left it largely to traditional industry associations to drive standards. In my opinion, it's been too slow. We've done integrations with 15 to 20 utility companies in the last 12 months, and every single one of them has been discrete and different. You just have to put the right team on it to know how to navigate the system and get through to the other end. But putting better standards in place would allow for what Mitch is describing. Your boat is charging your house when the power grid is down, or for four hours every afternoon when the grid is strained and demand charges are kicking in, or your trailer is charging your vehicle because it's got extra power. That's all interoperability that needs standards.
I would flag one other challenge. Mitch mentioned it, but it is one of our greatest pain points, which is supply chains. It's not just about getting the sell cost down so we can deliver cheaper product. It's being able to deliver product, full stop. The last couple of years have been really tough on supply chains. It's led to soaring costs, but also poor reliability. We're one of the largest buyers of medium and heavy duty truck platforms in an electric format right now. We're talking hundreds of millions of dollars per year of equipment purchases. We're not billions yet, but we will be quite soon. Even as a large buyer, it's tough to get product. You've got manufacturing sites that have thousands of completed vehicles, and they are all short an air compressor. So they can't complete these vehicles and deliver them to customers.
Guest 1
Preach. Yeah, we are buying switchgear 18 months in advance for projects we don't even have under contract yet. Because if we don't have switchgear, we're not getting the infrastructure completed and we can't get up and running. We don't get paid. We're solving problems that we shouldn't have to solve, in my opinion. I think for some entrepreneurs, there could be big opportunities in streamlining supply chains and creating better standards.
Guest 2
I heard a very creative description of the supply base over the past couple of years, and it was called the surprise chain. That's all you need to say.
Guest 1
Perfect. Yeah.
Guest 2
Timing for what we're doing. You look at the outlook on the aerospace industry right now, particularly aircraft manufacturing. A lot of the major aircraft manufacturers have already launched their next program. They're entering service, they're delivering them. If you take a look around at the large incumbent aircraft manufacturers, there's no new programs on the docket right now. The tier one supply base and even the tier two, tier three supply base, they're all hyper aware of what they're going to be working on. It gives us an opportunity to show, hey, we've got an aircraft, we've got a market, we've got good orders. The utility of the product is there.
I think what's going to happen is that the new technologies that we're bringing and our new products are actually going to garner the attention of the forward-looking supply chain. I think we're going to disproportionately benefit in the aviation industry, but I think the same is probably widely applicable. The new technologies, the new companies are actually going to benefit from obtaining the focus of the established tier one, tier two supply base. I don't want to say the problems are behind us yet, but I think the outlook is actually quite sunshiny.
Host
Another massive opportunity lies in the biggest industry in America. That's healthcare. This multi-trillion dollar industry, responsible for 20% of U.S. GDP and climbing, is widely referred to as broken. But there may be an opening to heal this problem with fintech. In this segment from our mini-series on A16Z's Healthcare Meets Fintech Package, we're joined by A16Z general partner Julie Yu, who exposes the huge opportunity at this intersection and how changing regulation has created an opening for new generation-defining companies.
Julie Yu
It's rare that people talk excitedly about regulation, but healthcare is one of those domains where I think regulation can be a tailwind for innovation and category creation. There's lots of historical examples of this. The most traditional example that a lot of people point to is electronic health records. They did not really exist in major adoption until the Meaningful Use Law came into play, where the government literally paid financial incentives to doctors to adopt digitized technologies for medical record storage. That was really the sea change that drove so much of the digitization of our infrastructure layer of healthcare.
Similarly, right now, we have a number of regulatory tailwinds that are driving payment related reform. We have things like a price transparency law that went into effect over the last couple of years that forced hospitals and insurance companies to publish their contracted rates. It was hugely controversial. There's still lawsuits in play. People are still pushing back. But the fact of the matter is, we now have thousands of hospitals and hundreds of payers who have published all this data. Obviously, they're publishing it in forms that make it very, very difficult to parse. Entire companies exist to actually process that data. We're investors in Turquoise Health, that's one of those players, and actually make it actionable in the context of their contract negotiations and the way that they engage with both providers and patients.
That's obviously a huge driver of change in terms of how we think about what used to be assumed to be opaque, now just being out there and people not having a place to hide when it comes to comparing prices between two different providers who are providing the same service, but at widely different prices. That's been a massive change that is only just starting to play out. I think we still have years for it to really seep through the system and address a lot of the challenges that we described earlier.
Host
You mentioned lawsuits and some pushback. Just because I feel like listeners might be curious, what is on the other side of that? It seems like maybe an obvious reform that we should be able to see how much something costs if we're paying for it or someone is paying for it for us. So what was on the other side of that?
Julie Yu
I have some degree of sympathy for these businesses. Effectively, what we're doing is taking proprietary contracts and publishing them on the web. There's been a lot of pushback from the parties to those contracts, which are the payers and the providers, who say it's proprietary data and we should not be forced to publish such data in a public forum because that's our competitive advantage in our market. We're able to negotiate special rates with our counterparties and we lose all of that competitive edge if we put it out there. That's the crux of most of the pushback from the incumbent lens.
The counter argument is that there had been previous laws that required things like upfront estimates for consumers. If you called your hospital, you were mandated to be able to present an estimate prior to coming in for a certain procedure or a certain set of services. From a consumer lens, those never really got implemented or enforced in a way that was reasonable, in my opinion. I remember my past life when that law went into effect in the state of Massachusetts, we actually did a bunch of secret shopper calls to hospitals to see what that user experience was like. You basically got told, "Oh, you'll get a call back in an undetermined amount of time." It was generally one monolithic number and there was no explanation as to what the range of assumptions that went into that number were. You get the call back like a week later when you might have had the procedure like two days after that call. The whole implementation was really poorly executed.
The argument on the consumer side is, listen, you guys promised this to us years ago. The form of that just didn't address any of our concerns about limiting the amount of financial exposure that we might have related to healthcare services. Those are the siren calls on both sides of the argument.
Host
So it sounds like we now have more transparency, but it also sounds like there were a couple other regulation changes that happened alongside that.
Julie Yu
Just maybe one more I'd call out, which we've talked about before, is all of the movement towards value-based payment models. This notion that a lot of the reason that healthcare business models were not resilient during the pandemic was that they were entirely fee-for-service oriented, meaning you only got paid for the specific services that you delivered. If the service was not delivered, you didn't get anything. Value-oriented models tend to be much more bundled in nature. Either getting a sort of a set upfront rate or price for a set of services related to an encounter.
A value-based orientation around a need for placement surgery would be, rather than charge for every individual doctor who's involved in the procedure, the anesthesiology, all the pre-visit, post-visit stuff, you'd actually just get charged one bundle for the entire journey. The provider is at risk. If they go above that budget, they are paying out of pocket effectively to cover the remaining cost. But if they stay under that budget and keep you out of the hospital and avoid errors, then they get to pocket the difference. It aligns incentives for the provider and the patient in terms of staying within certain bounds, but also creates a lot more resiliency around the payment flows into the provider practice because they're not just relying on individual services getting billed and paid for.
There's been a whole set of regulation around driving adoption of those payment models. Medicare Advantage, I think, has been the most prominent form of that. It's very early days in terms of adoption across our industry, but it is a very, very promising means to align incentives in a fundamentally different way that results in much more transparent behaviors.
Host
Absolutely. And maybe before we jump into the fintech side, just wanted to clarify the No Surprises Act and the Cures Act. It feels like those also were pretty fundamental to maybe changing this incentive structure that you've talked about so far?
Julie Yu
The No Surprises Act is very closely tied to the Transparency Act, which basically limits the ability to do surprise out-of-network billing for patients. You've probably read in lots of news articles these incidents where you go get a surgery in an in-network hospital, the actual surgeon is in-network, but the anesthesiologist is out-of-network. All of a sudden you're getting a bill for just that slice of the service at a rate that's outsized relative to what you would have had to pay had it been a network doctor as part of that procedure. This act limits the liability on the side of the patient and the provider in those situations.
The Cures Act is a little bit orthogonal to FinTech, but certainly related. The main provision that people care about is patient data access, enabling patients to readily access their full medical record data and limit the cost of access. It used to be the case that you'd have to potentially pay a couple hundred dollars for somebody to get access to your own medical records on paper, or even like a CD-ROM. This is a great example of an act that actually incentivized a whole bunch of startups to come out of the woodwork and create apps that allow you as a patient to basically collect all your medical records from multiple providers in a single swoop and really have agency over them.
The way that it relates to the fintech universe is that a lot of insurance adjudication acts require access to patient data to approve certain reimbursements. This act allows much more data liquidity to support those decisions versus the traditional way of doing it, which is faxing literally medical records back and forth and having nurses look at them in a manual fashion.
Host
And just as new regulation creates opportunity at the intersection of healthcare and fintech, new regulation is also driving pay transparency. With over 10 states now covering more than 25% of the U.S. labor force being mandated to disclose salary ranges for job listings, companies are now being forced to adapt, and adapt quickly. Up next is a clip from our episode on Pay Transparency with Shannon Schiltz, an operating partner leading A16Z's people team, and Brandon Cherry, a partner on the same team. They not only share how companies are reacting to changing regulation, but also touch on the importance of defining a consistent compensation philosophy.
Shannon Schiltz
Companies are now having to react to some of these changes in law. I'm just so curious to hear what you're seeing in terms of the reactions. Are they abiding? Are they struggling to catch up? Are they circumventing the system in different ways? How are you seeing companies react? Because it does sound like many were not quite in that place when these things started rolling out.
Brandon Cherry
I would say a couple things. One, we've probably never had so many requests for compensation councils, consultants, and "how do I?" People are reacting the right way. They just want to get their arms around it.
Shannon Schiltz
Yeah.
Brandon Cherry
There's multiple ways that people are reacting. One, some are just saying we're going to be in a wait and see. We're going to see how enforceable these laws are, how much audits are taking place, what becomes of this new law. There's companies that are saying we're not going to do it, so we're not going to post our jobs. Because that's the most visible thing that's come out of it, is you have to have a salary range posted with your jobs.
Shannon Schiltz
Yep, yep.
Brandon Cherry
And then the third piece is you have companies that are saying you could make between, this role pays $65,000 a year up to $2.5 million a year. Somewhere in there, we're going to pay someone. Those are probably the three things that we see companies doing.
Shannon Schiltz
Absolutely. Those are three choices companies have. There's nuance to it. In some states, there's not laws. There's minimum thresholds and things that people can explore around what the requirements are in different locations. I think there's just a lot of concern over making sure that not only do they understand how to comply from a regulatory standpoint, but also what does this do to our employee population? Who's going to see this? How is this going to work? If we do have $60,000 to $2.5 million, are we prepared? Do we have the ability to say, no, for you, it's $60,000. For other people, it's $2.5 million. How do you have that conversation?
Companies are navigating through all of that. For every company, it's going to be very specific to where they're located, the kinds of employee populations they have, how clear and how strong their infrastructure is, and how they feel like they're prepared to have those conversations.
Brandon Cherry
It goes back to the basics. If you have the structure, for us, it was relatively straightforward. We have the structure here. We could post roles and put a salary range out there. We set it up so we can defend it. The interesting thing that happens though is the number of times that somebody sees a job posted and thinks, "Well, that's what I do, but I don't make that." You have to be ready to have those conversations and get your managers ready to have a conversation on why actually this is a level above you and you can progress into it. But that's not where you're at right now. That's a hard conversation.
Shannon Schiltz
I was going to get into the hard conversations, but let's just talk about that now. Have you started to see that as you started to post roles?
Brandon Cherry
It helps us iterate on our process. No process is ever great. Weekly we're like, "Okay, are we checking this? Are we doing this?" We're iterating on it. Your future hire is super important, but they're not actually going to be contributing to productivity and output for six to nine months in most situations. You have to be able to defend it to your current employees. That's the biggest thing that people are missing. Yes, there is this law that you're supposed to post a range, but your biggest asset is your current employee population.
Shannon Schiltz
Isn't it true also that it's not just that current employees can see the new jobs, but they can also ask? What is the rule around that, that I, Staff Smith, can go and say, "Hey, I do this job." And what specifically can I ask for?
Brandon Cherry
You can ask for your pay range. That's where your manager should be able to come back to you and say, "Here's the pay range. Here's where you're slotted. This is where you came in. Here's how long you've been in this role. We're seeing the progression. Here's how we think about compensation." You would walk away thinking, "They've been super thoughtful in how I'm paid."
Shannon Schiltz
All right, let's start to think about how companies actually build this, because a lot of companies don't have this in place or are just in the early innings. If you are a company that has, let's say, the California threshold is 15 employees, and all 15 of those employees have been hired in disparate ways, paid in disparate ways, there's no sort of underlying strategy or structure. How do you even go about thinking about how to start setting that up?
Brandon Cherry
The three core tenets are what we've talked about before. You need to be able to articulate your philosophy on compensation. Not everybody is going to be able to compete with a $10 billion in revenue publicly traded company on a cash basis. You need to own the philosophy that you believe will drive the kinds of outcomes that you want as a company. Your compensation philosophy in terms of how you define your competitive market, companies of your size and scale, and where you position against that competitive market data is really important. That sets the tone for how you think about driving market data through your comp philosophy to determine those ranges.
The other element is your leveling architecture, how you define what it means to be an early career employee versus a seasoned individual contributor versus a first-time manager versus a seasoned leader. Between your philosophy and your leveling architecture and the third-party market data that you use to determine that relationship, you can set your ranges very consistent with what you should be providing to candidates and articulating to current employees when they request.
Host
Another area that experienced a renewed focus among entrepreneurs in 2023 was the world's largest asset class, housing. As the ripple effects from the pandemic have gradually played out, the housing industry has emerged as a space primed for disruption. Earlier this year, Mark Andreessen, co-founder of A16Z, sat down with A16Z general partner David Ulovich and Adam Neumann, co-founder of Flow and WeWork. Together, they highlighted some of the biggest opportunities in housing that still lie ahead.
Mark Andreessen
Why is housing important? Why is it particularly relevant? Why is it even a part of American dynamism? There's a critique of Silicon Valley in the tech industry that, on various states, sometimes I agree with, sometimes I don't. There's a critique that basically says, "Look, you guys have done a great job. Silicon Valley's done a great job over the years at the really unimportant things." Consumer electronics and video games and media and e-commerce. It's all great. It's all fine. It's all good. We do those things. We're very proud of our efforts in those spaces. We actually think they're very important, but they're basically small slices of GDP.
If you look at the GDP pie chart, the consumer spending pie charts, the things that people spend money on that really matter in their lives, there's basically three really big ones: healthcare, education, and housing. If you look at all the economic statistics by sector, healthcare, education, housing are basically increasing in cost at a much faster rate than all the others. They're the sectors where the tech industry has had the least impact historically. Of those, the biggest one is housing. Housing is the single biggest thing that most families spend money on. It's really central from that standpoint.
It's very central from an economic standpoint in the life of a family. There's this very big question: if you own, there are potentially big downsides because you might be stuck in one place where maybe you don't want your family to be for the next 30 years because opportunity shifts around. But if you rent, you don't necessarily build up equity by default with the way that most apartment offerings work. There's big economic consequences. There's also big social consequences and political consequences.
This is something that's been deep in the American character for many decades. This idea that if you have a sense of commitment, if you have a sense that you are actually invested in the place where you live, you are more connected to the society, you're more connected to the culture, and ultimately you're more connected to the political body. There's been long-running policies in the United States to encourage homeownership. It turns out those can be overdone. They can end up with problems. We saw a lot of the problems that came out of that approach in the 2008 financial crisis. Nevertheless, there is a very big, deep, meaningful sense of identity that comes with housing.
Why does this matter from a broad standpoint? Geography is so central to economic opportunities. Take a step back, the role of cities in human society. Cities are a human invention. There were many hundreds of thousands, millions of years where people were running around in clans and tribes, not forming into cities. About 4,000 years ago, cities were basically invented and people started to cluster that way. Why were cities invented? A bunch of reasons. But one of the big ones is there's economic payoff to being in a city.
If you have a certain level of productivity and you're in a rural environment, you don't have many opportunities to work with other people who are highly productive. If you move to a city, all of a sudden you're surrounded by lots of other people who are highly productive. All of a sudden there's a catalytic effect, a positive feedback loop that forms where everybody is productive. Cities are basically the foundation of modern economic growth. Economists call this the agglomeration effect: if you slam people together, you can get a lot of good things out of that.
The West developed. The story of the West is the story of cities and the creation of all these amazing cities. Over the last 50 years, that trend has gone into hyper drive. You've had this emergence of what economists call superstar cities, where you've got these specific cities like Washington, D.C., New York City, Boston, San Francisco Bay Area, Los Angeles, and internationally, London and Tokyo and Singapore and Paris. These are cities that are like irresistible draws for people who are highly ambitious and want to be around a lot of other highly ambitious, highly productive people.
As a result of that superstar city clustering thing, there's been a bifurcation in the housing market. Those superstar cities become occupied. At some point, people end up with a political agenda, and then they stop building housing.
And so the places where people want to go and want their kids to go are very difficult to get to. The San Francisco Bay Area is a classic example of this. San Francisco last year authorized, I think, 6,000 new housing units, which is an absurdly low number relative to the demand for people who want to move there.
The places that do not have this kind of clustering, agglomeration, superstar city effect build plenty of housing, but there's not as much opportunity there. So heading into COVID, you saw this incredible bifurcation in the American economy and in American politics. Basically, you're either in a city, you buy a home, you've got that old commitment, you've got that level of opportunity for you and your family. Or you're in a more rural environment, and you're basically stuck. You don't have access to good jobs, your kids are not going to have high-tech jobs, and you're going to have a lower quality of life.
This was becoming a very big issue for us, because we were basically getting stuck in the geographic confines of Silicon Valley. Our companies were penned in, and it was becoming very hard for companies to grow in that area.
And then a horrible tragedy struck that turned out to also be, in some ways, a miracle in terms of the long run consequences: COVID. None of us would have hoped that COVID would have happened. It has been a horrible tragedy for many people and for our society. But nevertheless, COVID was a system shock. It caused all big companies to basically instantly move online in a hard cutover, in a way that I never imagined was possible.
It was also an instant proof that companies really can run, at least for some time, online. No big company actually stopped operating as a consequence of this dramatic shift from in-person to online. All of a sudden, that opened the door to every CEO, every manager, every entrepreneur, every investor thinking, "Oh, okay, the post-COVID world is not going to be like the pre-COVID world."
The post-COVID world is going to be an opportunity to rethink and reinvent how companies are organized. It's going to be an opportunity to reinvent how industries are organized. It's going to be an opportunity to reinvent how geography works, what the role of cities is. It's going to be an opportunity to spread economic activity from cities much more broadly, potentially throughout the country.
Every employer, every company, every CEO here is having some version of this conversation with their own company. We talk about this with our founders all the time. There's a corresponding question that's opened up: how are people going to live? Some people are going to continue living the way that they were. Some people are going to undertake a radical change and go remote. All of a sudden, you go remote and you have access to thousands of jobs anytime you want, which is a totally new phenomenon.
You now have the opportunity to rethink how families live. Is it necessary for ambitious kids to leave a place where they grew up in order to have access to first-class economic opportunity? In 2019, the answer to that was frequently yes. Today, maybe not.
This is not a zero or 100% thing. This is not a hard cutover. It's not like the whole world goes remote or stays remote. But there's this moment in time that we're in right now where we can actually rethink and reinvent how companies are organized and how work happens. And we can also reinvent how people live and give people a lot more options in how that happens.
I think the presumptions that have underlined the whole structure of how the housing market works, how the industry works, and how people live and work are basically all up for grabs. I think we have about a five-year window as a society to figure out what this means and how to adapt.
Host
Finally, we hear from Steve Wozniak, one of the most important innovators and engineers of our generation. As you look toward the new year and think about how you want to get involved in this increasingly dynamic world around us, I hope you find Steve's reflections on inventors and entrepreneurs inspiring. One year at a time, people like Steve move us toward the future.
Another word sometimes people use for inventor is "visionary." I'm curious, in the early days when you were just out of passion creating these computers, could you see the path to today? Of course, you can't picture everything with so many advancements since those early days, but how far along were you actually envisioning?
I'm asking this partially because even if we apply this to space, a lot of the things that people talk about in the realm of space also sound kind of like science fiction. They probably won't be eventually. But I'm trying to understand how far along you see, or the extrapolation that goes on in your brain when you're originally talking about a computer with 200 transistors, and now we're talking billions and the applications that have sprung from that.
Steve Wozniak
I myself was really a great engineer in a certain field. I was designing the hottest products in the world for Hewlett Packard without even having a college degree yet. You talk about visionary, vision, seeing the future. That's different than invention, though. An inventor really wants to go in and create something today that didn't exist, and not have a vision that's 50 years out or 10 years out, because that's science fiction a lot. Everybody can talk about it and say later on, "See, I proposed it," but it wasn't possible to do with money.
The engineer says, "Feet on the ground. What can I actually do and build and deliver to people?" When we started Apple, we had a great product. It was going to be all the revenues of Apple for the first 10 years. We had a great lead. We were comfortable. We could do what we wanted. But the amount of memory that would hold a song cost a good fraction of a million dollars back in the days of tape. Do you think we saw it today where you have a device in your hand with a thousand songs on it? No.
Steve Jobs was very instrumental in always taking us to do what we can do today. Try to do something a little more tomorrow, a little more. You can have a lot of failures too if you have one great product bringing in the revenues. The whole idea was we'll move towards the future and we'll be a part of it and we'll be in with it. After all, you look back at it, it was kind of invisible, the steps we took, but they all led to today.
And then there was some of that invention stuff we got to. Steve Jobs' Apple II was really the iPod music. That was the first time — up till then our company valuation was the same as the old Apple II days. All of a sudden we sold it to everyone in the world and our sales doubled and our profits doubled and the board gave Steve Jobs stock options and jet airplanes. That was the turning point.
And then the iPhone was even better. It was based on the iPod, not the reverse. Not a phone that will include an iPod and more. It's an iPod, but you get a phone with it. So it's hard to say that you really see the future more than a year ahead when you're working a year ahead on your projects. Whenever I tried to see the future a year ahead, I knew it one year ahead because I was working on it. If I looked two years ahead and made some guesses, other aspects, other technologies came out from outer space and people's desire, which way they wanted to go was different. It's very hard to predict even two years ahead successfully the way I work.
Nowadays, we have huge big companies. It's kind of like anything they work on is going to be successful. It's not as much of a gamble. But real inventors like to gamble. They like to prove to the world that they can do more than you ever imagined.
Host
All right, that sums up our recap from the last year. Since our relaunch, we had over 75 amazing conversations and reached literally millions of people. It's truly been a blast and I cannot thank you enough for listening.
As we look toward the new year, this is your chance to let us know what you want to see us cover. You can leave your guest or topic selections in the comments on our YouTube channel, or you can even email us directly at [email protected]. You can also find the email in the show notes.
Again, I cannot thank you enough for listening. With that said, we'll see you bright and early in the new year.