Learning Library

← Back to Library

Gemini 3: The Next AI Reset

Key Points

  • Gemini 3, the first non‑OpenAI state‑of‑the‑art model, is set to trigger the biggest AI “reset” since ChatGPT’s 2022 launch, reshaping how consumers, builders, engineers, and executives operate.
  • The competitive landscape now hinges on five critical axes: frontier capability, default distribution, capital & compute resources, enterprise penetration/trust, and (implicitly) ecosystem integration.
  • Distribution advantage is key: Google embeds Gemini across Android (≈½ billion users), Apple relies on ChatGPT as its default AI app, Microsoft leans on Copilot in Windows/Office, while Anthropic remains a niche, non‑default option.
  • Capital dynamics differ sharply: OpenAI burns billions with profitability projected around 2030, whereas Google and Apple effectively have “infinite” cash for AI, and Anthropic is rapidly scaling to a multibillion‑dollar valuation but must manage frontier‑scale model costs.
  • Enterprise adoption and safety reputation are decisive: Anthropic already serves 300 k businesses with 80 % of revenue from enterprise, while OpenAI enjoys massive usage but faces heightened regulatory and trust challenges.

Sections

Full Transcript

# Gemini 3: The Next AI Reset **Source:** [https://www.youtube.com/watch?v=F-m4AIU8blY](https://www.youtube.com/watch?v=F-m4AIU8blY) **Duration:** 00:19:36 ## Summary - Gemini 3, the first non‑OpenAI state‑of‑the‑art model, is set to trigger the biggest AI “reset” since ChatGPT’s 2022 launch, reshaping how consumers, builders, engineers, and executives operate. - The competitive landscape now hinges on five critical axes: frontier capability, default distribution, capital & compute resources, enterprise penetration/trust, and (implicitly) ecosystem integration. - Distribution advantage is key: Google embeds Gemini across Android (≈½ billion users), Apple relies on ChatGPT as its default AI app, Microsoft leans on Copilot in Windows/Office, while Anthropic remains a niche, non‑default option. - Capital dynamics differ sharply: OpenAI burns billions with profitability projected around 2030, whereas Google and Apple effectively have “infinite” cash for AI, and Anthropic is rapidly scaling to a multibillion‑dollar valuation but must manage frontier‑scale model costs. - Enterprise adoption and safety reputation are decisive: Anthropic already serves 300 k businesses with 80 % of revenue from enterprise, while OpenAI enjoys massive usage but faces heightened regulatory and trust challenges. ## Sections - [00:00:00](https://www.youtube.com/watch?v=F-m4AIU8blY&t=0s) **Gemini 3 AI Paradigm Shift** - The speaker warns that Gemini 3, a non‑OpenAI state‑of‑the‑art model now embedded in Google’s platforms, will trigger the biggest AI reset since 2022, reshaping capabilities, distribution, and strategic priorities for users, developers, and executives alike. - [00:04:35](https://www.youtube.com/watch?v=F-m4AIU8blY&t=275s) **Gemini 3’s Potential Market Shift** - The speaker speculates that a breakthrough Gemini 3 model, if licensed to Apple and baked into both Android and iOS, could turn Google into the default AI engine for the world’s two biggest mobile platforms, but warns that Google’s slower production rollout and potential Apple‑imposed constraints could limit its distribution advantage, allowing competitors like OpenAI to retain market lead while Apple aims to leapfrog its own AI lag. - [00:07:52](https://www.youtube.com/watch?v=F-m4AIU8blY&t=472s) **Anthropic's Enterprise Surge vs OpenAI Challenges** - The speaker outlines hardware and UX setbacks limiting a new device’s rollout, warns that OpenAI faces a make‑or‑break need for massive scale or monopoly pricing, and highlights Anthropic’s rapid enterprise revenue growth and robust Claude model ecosystem as a contrasting success. - [00:13:03](https://www.youtube.com/watch?v=F-m4AIU8blY&t=783s) **Strategic Guidance for Enterprise AI** - The speaker advises enterprises to avoid relying on a single model, focus on integration surfaces, treat Anthropic as the safety benchmark, and monitor OpenAI’s burn rate as the market consolidates around a few major providers. - [00:16:07](https://www.youtube.com/watch?v=F-m4AIU8blY&t=967s) **AI Orchestration and Vendor Strategy** - The speaker urges professionals to shift from prompt‑crafting to mastering AI system orchestration—balancing cost, latency, quality, and security—while adopting a multi‑vendor portfolio and making explicit decisions between using OS defaults and building custom workflows. ## Full Transcript
0:00I believe we're headed into the most 0:01significant reset moment for AI since 0:042022 when chat GPT launched. Why is 0:07that? Because for the first time, we are 0:11about to see a new state-of-the-art 0:13model that has nothing to do with Open 0:16AI. That's Gemini 3, and it's going to 0:18change everything. I want to give you 0:20the strategic implications of that shift 0:24today. And I want to lay out for you the 0:26implications that you are going to see 0:28as a consumer, as an AI enthusiast, as a 0:30builder, as an engineer, and as an 0:33executive. I want you to think about 0:35this as a holistic shift in the 0:37landscape because I believe it's going 0:38to be the first thing we're going to 0:40talk about is the axes that matter in 0:42the board game ahead because I think 0:44that these are not widely understood. 0:46Number one, frontier capability is one 0:49of the five axes that matter. raw 0:50reasoning, how it does on benchmarks. 0:53We're actually pretty familiar with this 0:54one. I'm not going to take a lot of time 0:56on it. The key thing to remember is that 0:58for a long time now, OpenAI and Google 1:01and Enthropic have all been neck andneck 1:04around the top of the leaderboards and 1:06Chinese open- source models have been 1:08competing and just behind. Number two is 1:10distribution and who gets to have 1:12default status. Who owns the default 1:15surface for billions of users? Google 1:17has that on Android with Gemini 1:19integrated throughout. It's one reason 1:21why, and many people don't know this, 1:23there are half a billion Gemini users. 1:26Apple has this, but doesn't have any OS 1:28that actually is intelligent. And so, 1:30Chad GPT is functioning as Apple's 1:33default right now because that is the 1:35primary app that iPhone users are using. 1:38That's more vulnerable than you would 1:39think. Microsoft has a strangle hold via 1:42co-pilot on a lot of the Windows-driven 1:44office experience and anthropic is 1:46almost always in apps that you choose 1:49not defaults and that's going to come 1:51back. The third axis is capital and 1:54compute posture. So, OpenAI has a 1:57between a 12 and 20 billion revenue 1:59trajectory, but it's burning 8 to9 2:01billion a year and projected another $15 2:04billion of spend through 2029 with 2:07profitability not expected until 2030. 2:10Google and Apple, you can effectively 2:12thinking of them as having infinite cash 2:14for our purposes. From their core 2:16businesses, they're spinning off so much 2:17cash that AI is a line item. It is not 2:20an existential bet for them. I know that 2:22sounds crazy, but it is true for them 2:24both. anthropic. It's at five billion 2:27ARR in mid 2025 and it's scaling 2:29extremely rapidly. It will probably be 2:32valued at over $300 billion at its next 2:35raise. The capital question is not can 2:38they raise, it is can they sustain 2:41frontier scale model burn and keep unit 2:44economics somewhat sane for enterprise. 2:46Axis number four, enterprise penetration 2:48and trust. So Anthropic has over 300,000 2:52businesses now. 80% of their revenue is 2:54from enterprise and they have a very 2:56strong safety first brand that's helping 2:58them. Open AAI has massive usage high AR 3:01overall but is also kind of the poster 3:04child for regulatory scrutiny for the 3:06AGI risk and doom narratives. It has 3:08some brand issues. Google is a trusted 3:10infrastructure vendor for cloud already 3:13and has a long history of killing 3:15products and moving slowly which is not 3:16helping it in this situation. Apple 3:18maxes out the consumer trust axis but 3:21has minimal existing enterprise AI 3:23footprint and frankly minimal existing 3:25consumer AI footprint. The fifth axis 3:28around which everything moves is control 3:30of the UX layer. Whoever owns what you 3:33talk to wins a whole lot more than 3:36whoever owns the model. So Apple is 3:38trying to do this with Siri, but that's 3:40been a disaster. Amazon tried to come in 3:43and make a play for that with their 3:44inhome assistance. That's been a 3:46disaster. Google is trying this with 3:47Android voice. Open AAI has chat GPT as 3:50a voice but the voice model has not 3:52necessarily kept up with the pace of 5.1 3:55and the march of the models that are uh 3:58producing written text. Enthropic has 4:00web and API only and you can kind of 4:03compare it to having a strong brain but 4:06it doesn't have a lot of voice 4:07integration. The reset moment is that 4:09all five axes are about to move at once 4:12instead of one at a time which is what 4:14we've been seeing. Let's look at where 4:15the players sit on the board before we 4:17contemplate how the ball is about to 4:19spin. Google and Gemini from Lagard to 4:22OEM intelligence. So, their position now 4:24is at the frontier. Gemini 2.5 Pro is 4:27Google's top model. The company calls it 4:29the most powerful AI model today. You 4:30can make that claim, but whatever. It's 4:32it's in there. It has a lot of breadth. 4:34It has strong distribution, Android, 4:35Chrome, Workspace, etc. What changes 4:38with Gemini 3? If we assume, and it is 4:41an assumption cuz it's not out yet, but 4:42if we assume that Gemini 3 is a clearly 4:46accepted big step change 4:49state-of-the-art model, it is clearly 4:51better than everything out there today 4:53on all accepted benchmarks and a bunch 4:55of new ones. And if we assume it is 4:58integrated by default into Android and 5:01iOS via Apple licensing because Apple 5:04just cut a really big deal with Google, 5:06now we're in a different game because 5:08Google now shifts from being third 5:10contender in a race to the AI Intel 5:14inside for the world's two largest 5:16mobile platforms at once. Now, there are 5:20risks here. There are constraints here. 5:21If Apple wraps Jebana in its own UX and 5:24Apple wraps it in its privacy guarantees 5:26and Apple nerfs the model, Google risks 5:29being seen as just an engine, Google 5:31risks their brand and it may not happen 5:33very fast. Also, Google is historically 5:36slow at sort of productionalizing these 5:39research models. And so, it may be that 5:41we get Gemini 3 and it is incredibly 5:44good, but the distribution is not great 5:46and OpenAI is able to steal a march and 5:49keep their distribution advantage with 5:51the consumer. So, where is Apple? 5:53Apple's opportunity is really to move 5:55from AI lagger to potential leaprog. 5:57Apple's in-house models trail behind 5:59everybody else on metrics, obviously, 6:01but they're finalizing a deal to license 6:03a really big Gemini model for Siri and 6:06use that to power an Apple intelligence 6:08revamp. The cost is reported to be 6:10around a billion dollars a year. The 6:12plan would be to run a custom Gemini 6:14based model on Apple controlled cloud, 6:16keep the privacy narrative intact and 6:18use it to power a huge AI reboot for the 6:21company that would enable Apple to get 6:23Frontierra intelligence without eating 6:25the full capital expenditure of training 6:27Frontier models. And they can afford the 6:29cash, right? So they retain the OS 6:31integration, they retain all of the 6:33identity and the payment rails. They 6:35retain all of the hardware margins and 6:36the your data stays on your device 6:38story. If Gemini keeps pace or wins on 6:41quality, and if Apple can pull that 6:43intelligence in at a steady pace and 6:45refresh the experience so it stays 6:46cutting edge, Apple could leapfrog open 6:49AI on consumer UX, which none of us saw 6:51coming. Now, the risk is pretty simple. 6:53They're dependent on Google's road map. 6:55Any safety issues with Gemini become 6:57Apple's risk. Enterprise AI continues to 7:00be largely untouched by any of this. 7:02This is a consumer and ecosystem mode. 7:04It is not a cloud play. Meanwhile, if we 7:06go to OpenAI's side of the chessboard, 7:08they have very strong models at GPT5 is 7:11extremely strong on most benchmarks. It 7:13is the default mental model for AI for 7:15hundreds of millions of people. They've 7:17raised 40 billionish in capital. I keep 7:20turning around. They raise more 7:21billions, so who knows where they're at 7:22now. Their projected 2025 revenue is 7:24somewhere between 12 and 20 billion, 7:26give or take. And they're burning 7:28cashively. and they are trying to 7:30translate their cash into a cuttingedge 7:34frontier model position. So, OpenAI 7:38effectively bought Johnny IV's hardware 7:40startup to build a screenless AI device 7:42of some sort. That venture has 7:44reportedly hit technical and legal 7:46snags. A court has ordered them to pause 7:48marketing under the IO brand. 7:50Fundamental UX issues around how the 7:52device speaks and on device. There's 7:53leaks coming out of that team basically 7:55saying it's very difficult. Netnet, 7:57they're not shipping hardware yet. So 7:59now you're in a situation where you 8:00bought this device to help you to secure 8:03your advantage through this cash burn 8:05period, but it's just a lot of capital 8:07expenditure. There's a high uncertainty 8:09on the form factor. You haven't gotten 8:11to results yet. So to lad it up, Open AI 8:14is simultaneously a frontier model lab, 8:17a consumer app, and an infrastructure 8:19provider. They are in a go big or go die 8:21position. They need to either get to 8:23monopoly level pricing power which given 8:25the extreme proliferation of AI is 8:28unlikely or they have to go to extreme 8:30scale and multiple massive distribution 8:33partners maybe Microsoft maybe Apple 8:35OEMs whatever that's the only way they 8:38get to scale meanwhile anthropic quietly 8:40attacking the enterprise jugular they 8:42have scaled their revenue super fast 8:44they're on track for call it 9ish 8:46billion by the end of this year 20 to 26 8:49billion next year their valuation keeps 8:51exp- exploding and their base is 300,000 8:54plus business customers with largest 8:56accounts into six figures. So product 8:59stack claude models are very strong. 9:02They're near state-of-the-art. They're 9:03efficient. They're safe. The ecosystem 9:05is very strong thanks to the model 9:07context protocol adoption. Now thanks to 9:09Claude skills, uh claude code is very 9:12popular with developers. Almost all 9:14their revenue is enterprise unlike 9:15anyone else in this position. 9:17distribution is via platform enterprises 9:19already use AWS, Google cloud, direct 9:22API, SAS integrations. They have a very 9:25strong alignment first narrative which 9:26helps with enterprise focused on safety 9:29and they have economics that look much 9:30more disciplined than open AIs. 9:32Enthropic is essentially saying let open 9:35AI and Google fight over consumer. We 9:37will own the budget lines at the Fortune 9:39500. It might work. So here's what 9:41changes 9:43if Gemini 3 and Apple actually come 9:45together. we will move from a model arms 9:48race to a distribution duopoly on 9:51mobile. So instead of seeing a massive 9:53arms race across the whole spectrum, we 9:55will suddenly be in a world where Google 9:57powers the iOS experience by default, 9:59Google powers the Android experience by 10:01default and Google wins just about no 10:04matter what. We will also move from a 10:06world where we ask who has the best 10:08model all the time because they're so 10:10tightly competitive to a world where we 10:12ask who has the best UX and who has the 10:14best data loops because increasingly as 10:17models continue to get more effective. 10:19We're not going to be asking ourselves 10:21is the model smart enough to do it. 10:22We're going to be asking is the UX easy 10:24enough for me to use and is the data 10:27loop in place where I can get the data I 10:29need safely. A dumber model with better 10:31access to data is better today than any 10:33other model out there. It's also true 10:35that if the UX is terrible, you don't 10:37get the distribution. And that is 10:38actually the primary issue right now 10:40with Gemini is that Gemini's UX is not 10:44on par with where Claude and where 10:47OpenAI are. Gemini continues to be 10:50treated a little bit like a research 10:51project from Google and that is the 10:54historic risk of Google product 10:55thinking. I don't want to lose the 10:57narrative here either. If Gemini 3 is 10:59the clear state-of-the-art, Apple can 11:01credibly say, "We pick the best model." 11:03That makes it not a defensive choice 11:05anymore. Google will gain leverage 11:07versus AWS and Microsoft when they sell 11:09cloud AI because they can point to 11:11consumer dominance and they can point to 11:12state-of-the-art benchmarks and say they 11:14have the best. Open AI will lose some of 11:16their halo as the default synonym for AI 11:19unless they can deliver a model that 11:20beats. And this delivers a reset moment 11:22where the crown of best model and the 11:25crown of default assistant at once moves 11:27from OpenAI Microsoft to Google/Apple, 11:30at least in consumer. Now, if you layer 11:32in OpenAI's current trajectory and you 11:35look at their cash burn, this suddenly 11:37begins to matter strategically because 11:39if you're not obviously winning on 11:41distribution, spending tens of billions 11:43of dollars to stay at the frontier is 11:45going to be less defensible. Open AAI 11:47has a strategic imperative to continue 11:49to win at distribution and there is a 11:52real chance with the Gemini 3 moment 11:55that they will lose that edge. So what 11:58does this look like over the next couple 11:59years? If we fast forward, let's say 12:01Gemini 3 comes out, it's what we say 12:03Apple is able to move quickly. These are 12:04assumptions. They might not come true, 12:06but if they do and they're reasonable, 12:09then we have scenario A. Gemini is just 12:11everywhere. It's Google's winning all 12:12the way. Gemini 3 is the clear 12:14state-of-the-art. And then whatever 12:16comes after it, Apple's able to ship 12:18with the real brains of Gemini inside 12:20it. Anthropic eats enterprise share and 12:22OpenAI remains a strong player on web 12:24and app but loses default AI narrative 12:27and probably loses ground on enterprise 12:29and probably has some issues with 12:31fundraising down the road. Scenario B is 12:33a device reset. If OpenAI is able to 12:36ship a compelling AI native device, they 12:38could win the personal AI hardware 12:41subscription battle, harvest a ton of 12:43cash flow, and reset the bar for who is 12:46able to access AI as default and who is 12:48two hops away. Because if you can ship 12:50an AI native device, and it becomes the 12:53place where all of your voice is 12:54captured, now you're in a position to 12:56control the market. Scenario C is 12:59enterprise carve up and consumer chaos. 13:01The consumer space may continue to 13:03remain noisy with iOS having multiple 13:05players, Android having Gemini, multiple 13:08assistants, competing apps, etc. 13:09Enterprise buyers may consolidate on 13:12just anthropic and maybe OpenAI and 13:14maybe Google and just pick between them, 13:16which is a little bit like what I see 13:18today, and that might continue. If that 13:20happens, the winner is probably 13:22Anthropic because they thrive in a 13:24multimodel scenario. And the losers 13:26would be single model SAS vendors who 13:28have thin moes because this kind of 13:30carve up requires intense competition 13:33and thin moat SAS vendors are 13:36vulnerable. So what are the strategic 13:37implications here? Number one, stop 13:40treating your best model as your core 13:41bet. Assume that you need to swap 13:43models. I've said this before. I'm 13:45serious. Number two, optimize for 13:47surfaces. Don't just optimize for model 13:49IQ. Ask where does my user's intent 13:52originate? and then ask how can I build 13:54an opinionated workflow against that 13:56surface where they are against the voice 13:58against the Slack against the email not 14:00a generic chatbot if Apple and Gemini 14:03become the default assistant you'll want 14:05to design flows where you have that hot 14:07handoff from Siri or from Gemini into 14:09your app for specialized tasks number 14:12three start to treat Anthropic as the 14:14enterprise benchmark like take it 14:16seriously the way they invest in safety 14:18the way they invest in governance is 14:19something that I think sets expectations 14:21for a lot of production workloads. 14:23Number four, keep an eye on OpenAI's 14:25burn rate and keep an eye on the 14:27regulation and safety narrative at 14:28OpenAI. There are risks there. What can 14:31you expect to see changing depending on 14:33your role? If you're an individual, and 14:34I'll move up to executive from there. If 14:36you're an individual, you should expect 14:37your day-to-day tools will become more 14:39opinionated and more embedded. The idea 14:42of the best model is going to matter 14:43less to you than how you can orchestrate 14:45your tools around your work. And the 14:47half-life of specific tool skills is 14:49going to keep dropping. The half-life of 14:51judgment and the ability to design 14:53workflows is going to be very 14:54persistent. So optimize for how to think 14:57with AI, not a particular model. You 14:59want to treat your assistants like 15:00interchangeable contractors. And you 15:03want to become the person who can 15:04translate what leadership wants into 15:06what this stack of tools can do. If 15:08you're in the builder space, if you're a 15:10founder or a PM or a producty person, 15:12you cannot bet on a single model vendor 15:14or worthy assistant app as a strategy. 15:17Instead, you need to architect for model 15:19volatility. You need to pick a surface 15:20and obsess over owning it. Maybe it's 15:22spreadsheets, maybe it's email, maybe 15:24it's the terminal, maybe it's the 15:25calendar. Just own that. And then you 15:27need to differentiate on your workflow 15:29and on your proprietary data. So you 15:32have to have hard one process knowledge 15:34plus proprietary data or labels that are 15:36measurably better and then deliver into 15:38a domain specific UX that really adds 15:41value. Finally, financial discipline 15:42around AI usage is going to matter. You 15:44as a builder will have to make sure that 15:47usage can explode without token costs 15:49exploding. Your edge is going to be 15:51owning a specific workflow on a specific 15:53surface with a multimodel back end and a 15:56believable margin story. That's 15:57basically the big story you're going to 15:59have. If you're an engineer, the 16:01frontier model itself is less of a moat 16:03and how you use it is more of a moat. 16:05The stack is just going to get more 16:07complicated. So you need to start to 16:09learn to specialize in orchestration, to 16:11specialize in systems, not just in 16:12prompting. You need to design for tool 16:15and provider churn when you're thinking 16:16about your career and your systems. And 16:18you need to get really, really good at 16:20balancing cost, at balancing latency, 16:22and at balancing quality. Being able to 16:24show executives, we can cut 60% cost 16:27with 5% quality loss is going to be a 16:29big deal. And better or worse, security 16:31and data boundaries are part of your 16:32job. Now, you have to understand to 16:34things like tenant isolation, PII flows. 16:37You need to expect customers to ask 16:39about this and about which providers see 16:40their data and under what terms. If 16:42you're an engineer, your edge is going 16:43to be turning unstable models into 16:46stable systems that the business can bet 16:48on. If you're an you know seuite 16:50executive, if you own outcomes, budgets 16:53or teams, you are responsible for 16:56results, but you cannot pick the winner. 16:58You cannot pick a model. You have to 17:00adopt a portfolio vendor strategy. You 17:02need to plan on multiple primary model 17:04partners. You need to decide explicitly 17:07where you lean on OS defaults versus 17:09where you build your own. So if it's 17:11generic productivity, maybe the OS is 17:13okay. If it's core workflows for your 17:15business, maybe you invest in your own 17:17orchestration. But that line has to be 17:19an explicit choice. You need to start to 17:21frame AI as a workflow transformation, 17:24not software. I keep emphasizing this, 17:26but when you approve an AI initiative, 17:27the question is which workflow are we 17:29replatforming? What metrics will move? 17:32People can't answer that. If they can't 17:34talk about the workflow, they shouldn't 17:36be partners with you on that build. 17:38Governance and safety is going to be a 17:39bigger and bigger deal. We saw saw the 17:41anthropic hack this week. We need to 17:43have inventories of where models are 17:45used, policies on data residency, all of 17:47the stuff that goes with risk 17:49management. Also is going to become a 17:51sales enabler for you because other 17:53companies are going to take this serious 17:54too. On the talent side, you're going to 17:56need to be looking for AI native 17:57operators, not just prompt people. So 17:59the most valuable hires can map your P&L 18:01and ops to AI workflows, start to 18:03prioritize those by impact, and then 18:05work with technical teams to get them 18:07live. Titles are going to vary, but 18:09that's the capability that you're going 18:10to want to find. Last, but not least, 18:12you have to be disciplined with your 18:13capital allocation. Do not fund in-house 18:15model training, please, unless you have 18:17very clear reasons. Default to renting 18:19the intelligence and owning the data, 18:22the workflows, and the customers. 18:23Ultimately, your edge is going to be 18:24turning AI from scattered experiments 18:27into a coherent portfolio of bets that 18:29you can actually measure ROI against. 18:31This is what I want to leave you with. 18:33The strategic insight that we are on the 18:36verge of another reset, I think is 18:38stable even if the Gemini 3 and Apple 18:42story only becomes partially true. If 18:45Gemini 3 is a state-of-the-art model, 18:47but maybe not 20% better, maybe only 10% 18:50better, this could still happen. If 18:52Gemini 3 is embedded into Apple, but it 18:54takes 6 months instead of 3 months, this 18:57could still happen. The reason why is 18:59driven more by the strategic position of 19:02the players on the board where Anthropic 19:03is, where OpenAI is, where Google is, 19:06than it is by the exact timing of 19:08individual model releases. And that's 19:10what I think makes this a durable 19:11thesis. And I think it's worth paying 19:13attention to because we have not had a 19:15shakeup like this. We have not had a 19:17moment when OpenAI lost the crown. and 19:19we're about to find out what that looks 19:21like. So, get ready. The AI race is only 19:24heating up. I hope that this gives you a 19:26sense of where the market and where the 19:29AI space is going in general and what 19:32you can do to take advantage. Two.