Warren Buffett was asked in 1995 what P/E multiple makes a stock expensive. He refused to give a number. Berkshire had started buying Coca-Cola in 1988 at roughly $11 a share, and the stock did not look obviously cheap on current earnings.
By 1995, Coke was expected to earn around $2.30–$2.40 per share. Against Berkshire’s original purchase price, that was less than 5 times earnings. The original valuation did not change. The business grew into it.
Buffett’s point was that a P/E ratio only tells you what a company earns today. It says very little about what the business can earn 10 years from now, how much capital it will need to get there, or what return it can earn on that capital.
Interest rates matter too. A stream of future earnings is worth much more when long-term rates are 5% than when they are 11%. That is why there is no single P/E multiple that automatically makes a stock cheap or expensive.
A company trading at 8x earnings can still be expensive if profits go nowhere. A company trading at 25x can turn out cheap if earnings compound for a decade.
Buffett was not buying a multiple. He was buying the future economics of the business.
CHINA IS QUIETLY BUILDING ONE OF THE BIGGEST ROBOTICS BUSINESSES IN THE WORLD.
Take AGIBOT, a Chinese humanoid robotics startup. In 2023, the company made just $42K in revenue. One year later, it reached $8.4M. In 2025, revenue jumped to roughly $147M.
That's a 3,500x increase in just two years.
And they're not slowing down. AGIBOT has already produced 15,000 robots and is targeting around $1.4B in revenue for 2027.
The interesting part isn't just the robots. It's the business forming around them.
Factories, warehouses, hotels, entertainment and other industries will need these machines, and someone will have to sell them, integrate them, maintain them and find customers.
You don't have to build the robot.
YOU CAN BUILD THE BUSINESS AROUND IT.
Startup: AGIBOT
A warehouse robot doing its job is impressive, but nobody gets emotionally attached to it. Give a robot a face, a voice, personality and the ability to perform, and suddenly it's not just a machine anymore.
I can actually see people becoming fans of these robots. That's the crazy part.
🤖 CYBERPUNK IS ALREADY HERE. ARE POP IDOL ROBOTS THE FUTURE?
At the 2026 World Robot Conference in Beijing, AheadForm brought out Elf-Xuan 2.0 and had it perform live. Not a pre-rendered demo. Real-time lip sync, real-time facial expressions, arm gestures timed to the vocals.
Watch the face first. The mouth moves like it's actually singing. The eyes track. The expression shifts with the emotion of the line. Then watch the hand in the second clip, fingers articulating like a human wrist just decided to become machinery.
This is the same expo that showed off boxing robots and warehouse arms. Nobody argues in the comments about a warehouse arm.
They argue about this. Is it CGI. Is that a real robot. Would people pay to watch this on a stage, in a hotel lobby, as an actual idol?
That's the tell. Labor robots get specs. This gets a fanbase.
Twenty years of cyberpunk movies just walked off the screen and did a live vocal performance with perfect lip sync, on a conference floor in Beijing. If this is the prototype, pop idols might not stay human for long. 🤖
@notp3rk It's the fact that these agents can actually take action without waiting for you. If they can really handle tasks at this level for $2.81 each, I think we're getting very close to a world where you don't use AI tools anymore. You just give AI a goal and let it handle the work.
You don't need to own real estate to make money from it.
This is one of the simplest AI business models I've come across. You find property owners, help them market their properties with better AI-generated content, and connect them with potential clients.
The interesting part is how much of the work can be automated. AI can help create property videos, improve listings, find properties and even organize your outreach to owners.
You don't need to buy an apartment. You don't need a huge team. You don't even need to start with a lot of money.
You just need to find the right properties, offer something valuable to the owner, and take a cut when you bring the deal.
WATCH THE VIDEO BELOW. THIS IS HOW THE MODEL WORKS.
Everyone wants to make money with AI.
But AI itself is not the business.
The real game is finding a good niche where people already spend money, then using AI to do things faster, cheaper, or better.
You don't need a huge audience.
You don't need a million-dollar startup.
You just need one problem that people are willing to pay you to solve.
Find the niche.
Learn what people need.
Use AI to build the solution.
Get your first customers.
Then reinvest and grow.
Your first goal shouldn't be to build a huge company.
It should be to make your first $5,000.
And once you know how to do that, you can repeat the process in bigger and better niches.
RIGHT NOW, I'LL SHOW YOU A SIMPLE WAY I MADE MONEY BY RENTING APARTMENTS ON AIRBNB WITHOUT ACTUALLY OWNING THEM.
🤖 BEIJING JUST SHOWED WHAT THE NEXT ROBOT WAVE ACTUALLY LOOKS LIKE
The 2026 World Robot Conference just wrapped in Beijing. Aug 19 to 23. Over 300 exhibitors from 26 countries. More than 3,000 products on the floor. Over 300 of them debuting for the first time.
This wasn't a trade show. It was a preview.
Robots boxing. Robots dancing. Robots playing ping pong against actual humans. A 4-meter humanoid built for nuclear maintenance and deep-sea work. Robotic dogs trained for mine rescue.
And then the faces.
UBTECH showed off a humanoid built for emotional care. It talks, makes expressions, rotates its neck, reads a room. Price tag: 168,000 yuan, north of $24k. Not a factory tool. Not marketed like one either.
That's the split happening in real time. Some of these machines are built to work. Some are built to be looked at, talked to, trusted.
The opening day even coincided with Unitree's stock market debut in Shanghai. Shares jumped nearly six-fold. Investors aren't betting on robots that lift boxes anymore. They're betting on robots that hold your attention.
That's the actual headline from Beijing this week. Not "robots are getting smarter." Robots are getting sorted into two completely different products, and the market already knows which one it wants more of. 🤖