Is AI Too Big To Fail?
So much money has been poured into AI for Wall St, and the biggest tech companies have to make it work. What isn’t clear is what that means. Or is it?
It appears that the technology behind AI is advancing. It may not be as fast as had been expected. However, there remains a broad consensus that it is the most important technical advance in history. Engineers are expensive. Some have been offered several million dollars.
Distribution is an issue for everyone but OpenAI. ChatGPT has over 80% of the sector’s downloads. No other company has more than 10%. This holds for Google’s Gemini and Microsoft’s CoPilot, which is unusual given their size and penetration of their flagship products.
Microsoft And Meta
Downloads may not end up being the key metric. It may come down to how many companies are willing to pay for AI-based products. Enterprise and government sales will reach into the billions of dollars. And, that figure may be too low three to five years out.
It is the data center that requires real capital. And that is where institutions are putting their bets. Part of those bets is indirect, via investments in publicly traded market leaders. Meta and Microsoft are at the top of this list.
Despite the large war chests tech companies have, the cost of server farms will soon be close to a trillion dollars a year. This year, public company commitments to this part of the business are well above $500 billion.
Axios reports, “The promise of AI is sky-high, and with trillions of dollars, staggering energy use, and the markets behind it, so is the pressure to deliver.” That will be the case unless there is a train wreck. Right now, that is the electricity supply.
More from Electricity Wars
- Electricity Gets Ugly
- Another New York City Blackout
- Air Conditioning Takes Center Stage
- COP30 And The Rise Of Solar
