Wargames
One Of The Last Real Ones To Do It
The sad truth is the market is unstable and a bubble. Thiel wasn’t wrong to say “let me convert this shyt to cash”. He doesn’t need to risk money to make money.
Peter Thiel sells his Nvidia and Tesla shares. Sign of things to come?
- Thread starter Big Mountain Hélà
- Start date
Treblemaka
President, BYNKRadio.com (Retired)
Apple lost mobile device market share regardless. Adding AI to keep them from losing as much marketshare doesnt help them increase profitability, it just made the cut less deep. Thats not a ringing endorsement of a profitable application of AI.
Making strides in developing a product thats not going to become profitable for at least 5 years after market saturation (Waymo) at best doesnt change much especially when consumers are still very skiddish of self driving cars (and will be for awhile). Thats assuming the product is ready (which its not)
Waymo has received 30B in investment over the past 16 years primarily from Alphabet (another successful market leader in a different segment). Even if every dollar over that time had been invested in AI (which isnt even close to the case) thats still only a little more than 1% of whats needed to break even on 2 trillion in investments.
Do you know how many people would need to be paying for chatGPT at $20 a month each year to even make up half the 2 trillion in investment into AI at this point?
4.2 Billion people.
Literally most of the humans on the planet people would need to have a yearly chat GPT subscription just to cover 50% of current investments. This isnt even possible on today's Ai infrastructure and still wouldn't technically possible if there were another 2 trillion invested just on the infrastructure.
Do you see the problem here?
So back to the AI bubble concern and questions no one is answering:
What's the profitability model?
How do the trillions in investment get recouped?
Whats the answer breh? Im addressing all your points with tangibles and you havent addressed these questions with any hard numbers yet.
IIVI
Superstar
Some flaws:
You are conflating market share with profit share. Apple has never needed to dominate global market share to dominate global profits. They routinely capture over 80% of the entire smartphone industry's profits while holding a fraction of the volume. Using AI to reduce churn and keep users in that high-margin ecosystem is absolutely a profitable move. It protects the services revenue, which is their actual growth engine. If AI stops a user from switching to Android, that preserves thousands of dollars in lifetime value. That is not just damage control; that is defensive profitability.
Your take on Waymo is outdated. You claim the product isn't ready, but people are hailing driverless Waymos in San Francisco, Phoenix, and Los Angeles right now. It is a live, revenue-generating product, not a concept car. The skittishness you mention is fading rapidly in the cities where it actually operates because the convenience is winning. Comparing Waymo's specific investment history to the unrelated 2 trillion dollar figure for the entire generative AI industry is a logic error. Waymo doesn't have to pay back Nvidia's market cap; it just has to pay back its own fleet costs, which drop every year as hardware gets cheaper.
The math you did regarding 4.2 billion people paying for ChatGPT is a massive straw man argument. You are assuming the only way to monetize AI investment is through a single B2C subscription model. That is completely wrong. The 2 trillion dollars in infrastructure isn't debt; it is capital expenditure buying tangible assets like data centers and GPUs. Those assets are then rented out to thousands of enterprise companies via AWS, Azure, and Google Cloud.
The recoupment doesn't come from just ChatGPT subscriptions. It comes from Microsoft charging an extra $30 per seat for Copilot to millions of corporate users. It comes from Meta using AI to make their ad targeting 10% more effective, which generates billions in ad revenue. It comes from coding assistants making developers 30% more efficient, saving companies billions in labor costs.
The profitability model is efficiency and rental fees. The big tech companies are landlords renting out the digital infrastructure to every other business on earth. They don't need 4 billion consumers to pay $20 a month; they need the Fortune 500 to shift their IT and labor spend into AI compute. That is already happening. You are looking for a single killer app to pay the bill when the reality is that AI is a utility layer that taxes every industry that uses it.
Treblemaka
President, BYNKRadio.com (Retired)
Ai right now doesnt have a decision matrix that reduces churn enough to justify its expense. Apple lost 12% of its market share on its flagship product. Do you think the AI spend will reverse that enough to make it worthwhile investment?
Waymo has already had recalls on over 1500 vehicles this year alone. Liability accidents have increased over time not decreased (including liability for fatal accidents). Rolling out it pilot markets doesnt mean its ready for a national roll out (which they need to be profitable). There's a reason they are called test markets.
You brought chatGPT subs as a valid revenue stream. I showed the math on why that revenue is a drop in the bucket compared to whats required to invest, now its a straw man argument to you?
You mentioned $30 dollars a seat to corporations. The chat Gpt subs math shows why thats just another drop in the massive AI bucket called ROI.
CapEx is debt when you cant recoup it and these guys havent show how they'll recoup it.
"The profitability model is in rentals and efficiency"
Really? Where's the actual profit part of that? Where they part where they make more money than they are spending from it?
Rentals are drop in the bucket. Do you believe the efficiencies made are going to generate 2 Trillion in growth over the next 5-10 years? The current investment would require AI product companies are generating 200B-400B minus expenses every year over the next 5-10 to cover the current investment. To put that in perspective AI companies need to generate 50B more in revenue than Microsoft's total yearly operating expenses every year for the 10 years just to break even, and thats if another dollar is never invested. Do you believe the current AI ecosystem system is capable of that kind of growth because it hasn't show it yet.
Its clear from the math Seat licenses for an AI assistant arent a path to profitability. Total revenue generated from them doesnt even cover base infrastructure costs.
Im not looking for a single killer app, im showing you why every example you mentioned wont contribute to profitability enough to matter.
AI isnt going away. But the level of investment isnt justified if the expectation is a return. There's a bubble, the math doesnt justify it and it isnt sustainable.
Open AI is headed for a collapse and Sam Altman is reaching "found chopped up in the dumpster" territory
He can only do the "I just need 500 billion more mane" dream selling for so long
Exactly this is crypto hype all over again.
Promising technology, but is at the mercy of bad actors over selling and under delivering
UpAndComing
Veteran
Solar Energy was supposed to be the one to take out the Oil companies, then it was Wind Energy, then it was Electric Cars, now it's AI. When will people learn
Imagine going into your bunker thinking you're safe then not being able to get out once the smoke clears bc debris is covering your exit
IIVI
Superstar
You are looking at the Apple situation backward. The data shows Apple lost that market share specifically because they were behind on AI, not because AI is a waste of money. In China, where that drop was most severe, competitors like Huawei and Vivo were eating Apple's lunch precisely because they had advanced AI features that the iPhone lacked. Apple spending on AI isn't a gamble to fix a random decline; it is the required table stakes to stop the bleeding. If they don't spend that money, they lose more share, not less.
Regarding Waymo, you are misinterpreting what a recall means in modern tech. That recall of 600-1,200 vehicles was a voluntary over-the-air software update because a car hit a telephone pole at low speed. No one took their car to a mechanic; it was a software patch. As for safety, your claim about liability increasing is statistically incorrect. Insurance data from Swiss Re analyzed over 25 million Waymo miles and found they were over 85% less likely to cause bodily injury claims than human drivers. As they scale to more cities, the raw number of incidents naturally goes up, but the rate of accidents per mile remains far lower than yours or mine. They are already running 250,000 paid trips a week; that is a commercial business, not a test pilot.
On the financial side, saying CapEx is debt is a misunderstanding of accounting. CapEx is an asset on the balance sheet, like buying a house or a factory. Microsoft and Google are paying for this largely with cash on hand, not dangerous debt. You are also judging the return on investment on a timeline that doesn't make sense for infrastructure. These data centers and chips are assets that will operate for 10 to 15 years, but you are demanding they pay for themselves in year one of revenue. That is not how utility-scale investing works.
As for the revenue drop in the bucket, you are missing the cloud numbers. Microsoft's Azure revenue just grew massive percentages year-over-year, driven specifically by AI demand. That "rental" model you doubt is already generating tens of billions of dollars in real, realized revenue right now. Companies aren't just buying $30 seat licenses; they are renting the massive compute power to run their own internal apps. That is where the profit is. The efficiency gains for those companies, like developers writing code 40% faster, is why they keep paying the rent. The math works because the demand for that compute is currently higher than the supply, which is the exact opposite of a bubble that is about to burst.
Also funny part about this is my last few replies have been using A.I, literally only spent maybe 10 minutes total (not each message) with these replies today. I’m just copying and pasting with a phone.
Last edited:
Treblemaka
President, BYNKRadio.com (Retired)
You are looking at the Apple situation backward. The data shows Apple lost that market share specifically because they were behind on AI, not because AI is a waste of money. In China, where that drop was most severe, competitors like Huawei and Vivo were eating Apple's lunch precisely because they had advanced AI features that the iPhone lacked. Apple spending on AI isn't a gamble to fix a random decline; it is the required table stakes to stop the bleeding. If they don't spend that money, they lose more share, not less.
Regarding Waymo, you are misinterpreting what a recall means in modern tech. That recall of 600-1,200 vehicles was a voluntary over-the-air software update because a car hit a telephone pole at low speed. No one took their car to a mechanic; it was a software patch. As for safety, your claim about liability increasing is statistically incorrect. Insurance data from Swiss Re analyzed over 25 million Waymo miles and found they were over 85% less likely to cause bodily injury claims than human drivers. As they scale to more cities, the raw number of incidents naturally goes up, but the rate of accidents per mile remains far lower than yours or mine. They are already running 250,000 paid trips a week; that is a commercial business, not a test pilot.
On the financial side, saying CapEx is debt is a misunderstanding of accounting. CapEx is an asset on the balance sheet, like buying a house or a factory. Microsoft and Google are paying for this largely with cash on hand, not dangerous debt. You are also judging the return on investment on a timeline that doesn't make sense for infrastructure. These data centers and chips are assets that will operate for 10 to 15 years, but you are demanding they pay for themselves in year one of revenue. That is not how utility-scale investing works.
As for the revenue drop in the bucket, you are missing the cloud numbers. Microsoft's Azure revenue just grew massive percentages year-over-year, driven specifically by AI demand. That "rental" model you doubt is already generating tens of billions of dollars in real, realized revenue right now. Companies aren't just buying $30 seat licenses; they are renting the massive compute power to run their own internal apps. That is where the profit is. The efficiency gains for those companies, like developers writing code 40% faster, is why they keep paying the rent. The math works because the demand for that compute is currently higher than the supply, which is the exact opposite of a bubble that is about to burst.
Also funny part about this is most of these replies have been using A.I, literally only spent maybe 10 minutes total (not each message) with these replies today.
Accidents are increasing is what I was saying thats a facts. I didnt say anything about mechanics I said they recalled a product. Thats a fact.
The difference with something being safer than a human doesnt matter to a human when it still causing accidents. People dont use calculators because they are 85% better than a human they use them because they are perfectly accurate compared to a human. People dont automate factories because they are 85% more efficient, they do so because its 1000s of time cheaper an infinitely more reliable.
If the product is so stable why haven't they started a nation wide rollout?
Azure was already profitable, established and growing when AI was added. Youre talking about 10s of billions generated when this needs 100s of Billions generated yearly to break even.
A factory has a immediate purpose and industrial value because of the products it is specifically designed to make in a way that dramatically increases profitability due to scale. Which product when adding AI makes it drastically more profitable due to scale? Because it aint Azure.
Those datacenters will have to replace GPUs every 3 years by Nvidias own calculation and every 5 by the most liberal estimation of chip life. Where are you getting 10-15 from? Please name a product GPU that can last 10 years being redlined the way these chips are.
The chip manufacturers arent AI companies they are chip manufacturers. They will make money regardless of the aI bubnle and they are selling more chips to AI companies that dont have a profitability model and dont have a path to be. This isnt different than building a bunch if malls to lease to stores for that wont be there long because they arent making money. This means investment will slow eventually and the chip OEMs banking on it are going to decrease in value when it does.
Youre using Ai but its not helping your argument. It has access to all the world information but nothing you've said is a viable path to profitability. You've even mentioned grossly inaccurate things in relation to this conversation.
But if it saved you time I guess thats what counts. Right?
Last edited:
IIVI
Superstar
You are looking at the accident data without context. Yes, the raw number of accidents has increased, but that is simply because the number of miles driven has skyrocketed. If you drive one mile and crash zero times, then drive a million miles and crash once, your accidents "increased" but your safety record is still phenomenal. The metric that actually matters for profitability and insurance is the accident rate per mile, which remains consistently lower than human drivers.
As for your calculator analogy, it doesn't apply to transportation because the alternative isn't a perfect machine, it is a flawed human. We do not need self-driving cars to be perfect like a calculator; we need them to be statistically safer and cheaper than a human driver to make the business model work. If a computer drives a taxi for half the price of a human and crashes half as often, the market will switch to the computer regardless of whether it is perfect[1]. The reason they haven't rolled out nationwide is regulatory and logistical, not technical failure. You can't just download a car; you have to map cities and get permits, which takes time. [2][3][4][5]
Regarding the hardware lifespan, you are conflating the chips with the infrastructure. The 10 to 15-year lifespan applies to the data centers themselves, the concrete, cooling systems, power delivery, and networking fiber that make up a huge chunk of that CapEx. You are right that GPUs get cycled out every 3 to 5 years, but they don't go in the trash. They get moved from high-end training tasks to lower-end inference tasks, extending their economic life.
On the profitability point, you are ignoring that the companies spending this money are already making the money back through the cloud. You ask what product makes this profitable, but the answer is right in front of you. It is the compute capacity itself. Microsoft and Google are acting as the landlords. They rent the "factory" (the GPUs) to thousands of other companies who use them to build their own products[6][7]. You say AI companies need to generate hundreds of billions to break even, but the biggest spenders are Meta, Microsoft, and Google, who essentially have infinite runways and are funding this with cash from their existing monopolies, not debt that will bankrupt them.
Finally, saying the efficiency gains are a drop in the bucket ignores how corporate budgeting works. If an AI tool lets a bank automate their customer service center or helps their developers write code 30% faster, that bank will pay millions for that software because it saves them billions in labor costs. That is the path to the trillions in value you are questioning. It isn't about one killer app; it is about shaving distinct percentage points off the cost of labor across the entire global economy. That is where the ROI lives.
Sources help
Last edited:
Treblemaka
President, BYNKRadio.com (Retired)
Im going to assume these are more AI answers.
Tell the average human being to get into a car that still gets into wrecks but wrecks less than the average person and watch how fast they get back in their own car and drive off. Its even more critical in transportation because a calculator cant kill you. A failed sensor on a waymo vehicle can.
The most expensive part of datacenters are the power and GPU costs. Those cost dont decrease over time they increase over time. And according to the chip manufacturer (Nvidia) 3 years is the total device lifecycle. Itd be like buying a used GPU from a bitcoin mine. They are practically shot, they dont usually get cycled to lower computing requirements they get trashed.
Microsoft cant outspend its entire operating expense sheet to fund AI and it isnt going to, neither is Meta, or Google because they don't have infinite revenue streams they just have Hundreds of Billions in revenue and a high market cap. And they arent spending the majority of that money on AI. Even if all three of those companies spent 10 percent of revenue on AI for the next 10 years combined it wouldn't even equal 40% of the required spend yearly earnings required to break even on current investment in 10 years.
Breakdown the customer service center example. What do you think it costs to maintain a tool that allows the elimination a call center that is likely already offshore and costs pennies on the dollar to maintain compared to an American one.
How much does the product cost?
Whats the year over cost savings?
What is the profit margin of the product?
How many banks are likely to buy the product annually?
What are the annual product maintenance costs?
Pretend youre the Ai company and show me how this becomes profitable based on the startup spend required.
Im trying to understand the details.
Last edited:
IIVI
Superstar
Funny thing is, that’s why be if the things we’re doing at my company: building an IVR that connects to A.I for things like a call center.
Nonetheless:
The data on Waymo adoption proves that your theory about human psychology and fear is incorrect. You claimed people would not get back into a car that is not perfect, but Waymo is currently completing over 150,000 paid trips every single week. That number has tripled in just a few months. If the fear of a sensor failure were stopping adoption, you would see empty cars, but instead, they are operating with waitlists in San Francisco, Los Angeles, and Phoenix. The general public has voted with their wallets that a car that crashes significantly less often than a human is safe enough to use, even if it is not theoretically perfect.
Regarding the data center economics, you are looking at the total cost rather than the unit cost. While the total power bill for a data center rises, the cost to produce intelligence drops dramatically. For example, the transition from Nvidia’s H100 to the Blackwell chip increases performance per watt by roughly 25 times for inference tasks. This means that while the hardware is expensive, the cost to process a single transaction or customer query decreases over time. This is a standard deflationary technology curve, not a cost spiral that makes profit impossible.
On the financial capacity of Big Tech, your math regarding their spending limits does not align with their balance sheets. Microsoft generated over 118 billion dollars in operating cash flow in fiscal year 2024 alone. They are projected to spend about 50 to 60 billion dollars on capital expenditures. They are paying for this infrastructure entirely with their own cash while still having tens of billions left over for dividends and buybacks. They do not need 200 to 400 billion dollars in profit immediately to break even because these are long-term assets that depreciate over 5 to 7 years, not expenses that vanish in a single quarter.
For the customer service breakdown, the math is quite specific. You mentioned offshore agents cost pennies, but that is an exaggeration. A low-cost offshore agent costs a company between 5 and 10 dollars per hour once you factor in training, management, and infrastructure. If that agent handles four tickets an hour, the cost is roughly 2 dollars per ticket. In contrast, a generative AI solution costs fractions of a cent per token. A full resolution of a complex customer issue via AI costs roughly 20 to 30 cents in computing power.
We can look at the Klarna case study for the exact profit numbers you asked for. Klarna released data showing their AI assistant handled 2.3 million conversations in its first month, which is the equivalent workload of 700 full-time agents. They stated clearly that this implementation is expected to drive 40 million dollars in profit improvement in 2024 alone. The cost of the product is the API usage paid to OpenAI, which is a variable cost that is significantly lower than the fixed cost of human labor. The maintenance cost involves a small team of engineers managing the prompts and data integration, which is far cheaper than the HR and management costs of maintaining a 700-person call center.
The path to profitability for the AI company comes from volume and margins. The AI provider, like OpenAI or Microsoft, sells access to these models at a gross margin estimated between 50 and 70 percent. Because the cost difference between a human agent at 2 dollars per ticket and an AI agent at 30 cents per ticket is so large, banks and retailers are buying the product in massive volumes. The AI company becomes profitable by selling billions of these low-cost transactions, covering their high startup spend through the sheer scale of global enterprise adoption.
Last edited:
