# AI capex — X 热门讨论 (2026-09-16 07:21 UTC)
## @WendySol_ (🌱 Wendy) · 09-16 06:30 · ♥40 ↻4 💬35 Good Morning & Happy Hump Day!
Your smartphone might be more useful to AI than you think.
It already has cameras, motion sensors and the ability to capture what is happening around you.
@vangrid_io is taking that existing hardware and thinking much bigger:
what if everyday devices could become part of a massive, distributed spatial data network?
That creates a path toward continuous real world ground truth without needing to deploy dedicated sensors everywhere.
And ground truth matters.
Robots need to understand their surroundings. Autonomous systems need current spatial information. Physical AI needs more than text and images sitting in a dataset.
It needs the world as it actually is.
Vangrid describes this as a zero capex sensor swarm, built around decentralized scale, multi view ingestion and cryptographic provenance.
So the smartphone isn't the end product.
It's one of the capture points for a much larger spatial intelligence layer.
That shift from isolated devices to a collective view of the physical world is what makes Vangrid worth watching. 📱🌍 https://x.com/WendySol_/status/2100109932557463729
## @geniusparadox (Sanjeev) · 09-16 05:26 · ♥38 ↻10 💬4 I recently had this discussion with one of my friends. He is in IT and is natually a bit worried about his future because of the disruption caused by AI. He called me seeking my opinion on him entering copper industry by installing a mill. He wanted to import scrap and process them into finished products. He spoke very excitedly about the whole thing as in his mind he had imagined the whole thing out. Running big numbers in and bigger numbers out.
Then i explained to him in detail on what the project will entail. The capex required, the working capital required, the statutory approval required, the risk of importing scrap especially copper, the labour problem involved, and most importantly finding the right product, selling to the right customers buying at the right price and them paying for it (its a bonus if its on-time). He immediately realised the enormity of the challenge and most importantly risk that it involves. Its one thing to see your 50L portfolio on zerodha bleed 20% and have *unrealised loss* of 10L, and an entirely different thing to have rejections because of an operator’s negligence, or have goods stolen or simply customer not paying and have an immediate *realised loss* of 10L with no recourse. And to live in this risk every day of every month of every year that you are in manufacturing. Add to this geopolitical risks which can affect your input costs and can undo work of 9 months in just 3 months and wash all your margins away. Add to the mix some gst-customs officers harassment and it’s a life nobody would want.
I am not painting a pessimistic picture. There is a lot of satisfaction in being a manufacturer. The high of finding a customer and having repeated business with him whilst most of your machines are running well, to full capacity and are making money for you is unparalleled. To develop a new product from scratch and see it in reality is pure happiness. To see people using something you produced is unencumbered satisfaction! But it all comes at the costs of what is explained above.
To everyone who wants to enter manufacturing- first try to find a product you want to sell. At the end of the day, the objective is to sell. Once you have found that product, do extensive market research on it. Then start trading it. Your risk will be limited to your working capital and to handling the stock only. Find your first customer, then group of customers. Find some stability. Then slowly start backward integrating. You will then find a purpose for your manufacturing. You will by then have clarity on your margins and how having a manufacturing setup will improve those margins. Then you scale. In manufacturing growth accelerates with growth. I started my year 1 with 1 machine. It took my 2 years to add my second machine. Today in year 4 i have 7.
A more sensible approach would be to first work in the area of manufacturing you want to be in. Work as an employee. Get to know what the business is. What the risks are, what the rewards are. So that when you invest your own capital, there are lesser surprises and fewer learning from your own mistakes. And be prepared for slow organic growth. I am a very strong believer in sustainable organic growth because it gives you better control over the steering of the company.
I really want the younger generation to take to entrepreneurship and manufacturing better than my generation did. But please do so while taking cautious steps. There are many instances where promising entrepreneurs full of potential have been lost because of the crushing defeat manufacturing and entrepreneurship hands you if you take even a few wrong steps. India needs manufacturing, more so now than ever. Heres me praying and wishing that children of Bharat maatha rise up to take risks and make her proud! > 引用 @prakdadlani: My manufacturing posts are probably giving you the wrong idea.
I get DMs and calls every week from people who want to become manufacturers.
I get it.
My posts make this world look sexy.
Trust me - it isn’t.
Manufacturing needs sales, capital, relationships, patience, problem-solving and, most importantly, the willingness to learn the hard way.
You can’t watch a few videos, take a few calls and think you understand the game.
You have to get your hands dirty.
Talk to buyers, visit factories, lose money, fix problems and start all over again.
Passion gets you started but skin in the game makes you good.
It’s a grind.
And you have to genuinely want it. https://x.com/geniusparadox/status/2100093833124172139
## @KingMancer (KING MANCER) · 09-16 04:32 · ♥41 ↻1 💬5 The FOMC decision isn't the whole story for semis. The real question is whether the Fed gives clarity on where the tightening cycle peaks.
Once that's clear, attention shifts to AI capex, data centers, and chip demand. That's the part I'm tracking on BingX TradFi.
#FOMC https://t.co/DIxOhbEsRA https://x.com/KingMancer/status/2100080234846425186
## @FindleysFinance (FindleysFinance) · 09-16 00:56 · ♥35 ↻3 💬1 Bill Ackman is spot on. The market is starting to understand the difference between growth and maintenance capex.
$META spending is clearly paying off with revenue growth of 28% and OCF growth of 24% YoY.
Don't let AI doom slow us down. Let them build! https://t.co/qVZX5lfSrb https://x.com/FindleysFinance/status/2100025919360381058
## @KrisPatel99 (Kris Patel 🇺🇸) · 09-15 18:18 · ♥32 ↻3 💬4 This is a great comment.
He accurately points out that the current rate of return for SP500 is 2.5% while long term yields are 5%+.
The thing that’s keeping equities up is the growth rate. Earnings are growing and compounding faster… for now…
Most of SP500 index is now concentrated into a handful of names that are growth earnings at a brisk pace. That’s keeping the entire index afloat. Debt fueled AI capex is spending is driving liquidity into those names and pushing the growth rate. That same debt issuance is competing with long term treasury yields.
The cycle is strong right now but any weakness or credit deterioration in the underlying issuers of the capex debt or even a slow down in growth will start hitting forward market multiples.
This will then push people out of equities and into bonds. > 引用 @InvestiBrew: @Mr_Derivatives When you look at the earnings yield for the $SPY $SPX being lower than 2.5% and a $TLT yield of 5%...
It makes no sense to still see stocks trade at the high point for the year anymore https://x.com/KrisPatel99/status/2099925918634426476