The very thing driving the race to AGI, money could also ultimately stop it. AI companies are chasing AGI for profits, but if stocks bleed and investor confidence collapses, the AI arms race could come to a halt.
Maybe in the long run this is better alternative than skynet.
For years, Google was one of the biggest buyers of Google stock. This fall it stopped buying and started selling brand-new shares into the market instead, and hedge funds ran for the exit.
The swing is huge. Alphabet bought back $11.5 billion of its own stock in the third quarter of last year. In the same quarter this year it bought back zero and sold $12.4 billion of new shares.
A company quietly buying its own stock every quarter is a steady buyer propping up the price. Once that buyer turns into a seller, the stock trades a lot differently.
Meta just did the same thing. Five days after reporting its first negative free cash flow quarter in more than a decade, it launched a stock sale that can reach $97.5 billion, including a $60 billion program to drip new shares into the market starting early next year. Nvidia put in $3 billion, so a chip supplier is now funding its customer so the customer can keep buying its chips. Add the $84.75 billion Alphabet lined up in June and these two companies have more than $180 billion of new stock coming in five months.
The funds that owned the most felt it first, and they owned a lot. Even after the selling, tech was still 23.5% of Goldman's hedge fund book, higher than about nine out of ten readings over the last five years. In the five trading days to October 15, funds cut tech faster than at any point since Goldman's records start in 2016. A week later, Goldman's basket of hedge funds' favorite stocks fell 4.2% in a single day, its worst since 2011. When everybody piles into the same handful of names, they all hit the same exit at the same time.
Earnings week made it worse. Alphabet had the worst day in its history as a public company and lost about $489 billion in value in one session. Intel fell 29% on Friday, its worst day since 1974.
Where the money went is the part people are missing. On October 29, 406 of the 500 stocks in the S&P went up and the index still closed lower, because Alphabet dragged it down by itself. Bloomberg's data back to 1996 had never seen that happen. UBS traders say the money is moving into value stocks, small caps, consumer names and rate-sensitive cyclicals, so it's staying in the market and just leaving the AI giants.
For years those companies paid shareholders to stick around with buybacks, and this fall they started asking shareholders to help pay for the data centers, which is right when the biggest holders headed for the door.
You know you're dealing with a great bank when the charges are auto deducted but to reverse the same you need to visit the branch and beg them for it.
@aubankindia
@aubankindia Auto-deducting charges is easy, but reversing them requires a branch visit? Did the branch visit me before deducting my money? Pathetic customer service. Taking money is automatic, returning it is a struggle!
Code in Claude, ask Codex to check for issues its finds 4. Ask Claude to reverify and it again finds 5 more.
Its a never ending vicious circle. Code can never be perfect and bugs will always exist
I just sent a single prompt to Claude and it instantly used my 5 hour limit, with just one prompt and within 1 second.
The prompt was "Continue"
Why does Android still get Whatsapp calls wrong?
On iPhone, if I’m already on a regular cellular call and someone calls me on WhatsApp, it’s handled quietly in the background.
On Android, WhatsApp can literally start ringing loudly in my ear while I’m already talking to someone.
Surely this can be fixed. iOS has had the experience right for years.
Gemini has the worst coding capability. Any file it touches, it will overwrite huge sections of code and then spend half an hour trying to piece it all together.
If it doesn't destroy existing code, the new code it writes is so poor that it just won't run.
ChatGPT Codex and Claude win hands down. The code quality is superb and even if there are errors, they are small and easily fixable
Using AI to code is like getting addicted to games. No matter where you want to stop, there is always that one feature that seems worth adding.
The list just goes on and on even for apps which have zero users in real but the features seems like a no brainer.
Ai models will build you your app in a few prompts, but the underlying architecture is a different story.
If you don't tell it specifically about the security, implementation best practices and industry standards you will have an app, but that app will be a toy at best.
Something is seriously wrong with @GeminiApp Asking it to even fix a simple issue results in massive lines of unrelated code being deleted & overwritten. It consumes 2% of the weekly usage to write the fix and 98% to rewrite the unrelated code Gemini Flash 3.6 destroys
Even after all these years, I still don’t understand why Amex can’t make card payments seamless.
Why do I have to go to a separate website, enter my full card number, outstanding amount and other details just to pay my Amex bill?
Why can’t this be integrated directly into the main Amex website/app? Seems like such a basic thing to fix.
@AmericanExpress
The day a blind man sees, the first thing he throws away is the stick that has helped him all his life - African proverb / anonymous
There’s something so charming about these old Delhi houses that still feel frozen in the ’60s and ’70s — a little verandah, a grilled gate, and just enough space to park a scooter or cycle. Weathered walls, moss, faded colours, and so much character. They don’t make homes like this anymore.
