Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apparently companies are reluctant to build new Fabs, as they're betting this is a bubble and setting up new capacity is very expensive. These companies tend to follow a more long term planning, due to the capital intensive / high risk nature of the industry.
Why not ask Macalway AI?


Is ai a bubble ?

The question of whether AI is a "bubble" is currently one of the most debated topics in economics and technology. To determine if we are in one, it helps to look at the classic characteristics of a financial bubble: massive capital inflow, sky-high valuations, and "irrational exuberance."

Here is a breakdown of the arguments on both sides:

The Case for "Yes, it's a Bubble"​

Critics argue that the current AI boom mirrors the Dot-com bubble of the late 1990s.

  • Infrastructure vs. Revenue: Companies are spending billions on "shovels" (Nvidia chips and data centers), but the "gold" (profitable AI applications for consumers) hasn't fully materialized yet.
  • The "AI Premium": Many startups are seeing massive valuations just by adding "AI" to their pitch decks, even if their underlying product is just a wrapper for existing models.
  • High Energy & Computing Costs: The physical cost of running these models is immense. If companies can’t find a way to make the profit exceed the electricity bill, the investment might dry up.

The Case for "No, it's a Fundamental Shift"​

Proponents argue that unlike the pets.com era, AI is already providing tangible utility.

  • Immediate Productivity: Unlike the early internet, which took years to become useful for business, AI is already being used to write code, automate customer service, and speed up drug discovery.
  • The "Gartner Hype Cycle": We might be at the "Peak of Inflated Expectations." Even if a crash happens, the underlying technology usually survives and becomes the foundation for the next era of the economy.
  • Gartner Hype Cycle diagram showing the peak of inflated expectations and the plateau of productivity, AI generated

    Getty Images
    Explore

What happens if it "pops"?​

If the bubble pops, it doesn't mean AI disappears. Usually, a "pop" looks like:

  1. Consolidation: The hundreds of "AI startups" fail, leaving only the giants (Google, Microsoft, OpenAI) and a few specialized winners.
  2. Reality Check: Stock prices for chipmakers and tech firms return to more modest levels.
  3. The Boring Phase: AI stops being a flashy headline and simply becomes a standard tool integrated into every piece of software, much like databases or high-speed internet.

Where do you see the most impact? Are you seeing AI actually change how you work, or does it feel more like marketing noise right now?
 
  • Like
Reactions: drunkaviator
Blessing in disguise perhaps. Time to de-bloat all of the apps to use RAM more efficiently. Surely, AI could help achieve that?
Who optimizes these days, don't be silly 😉
I'm helping a client with some tech bloat, a whole lot of servers with 50-100% more memory than they need, all increased due to lack to tuning. My AI created tool kit find leaks and issues super quickly and we're getting so many things working better. AI can definitely help.
 
Prices of devices will definitely increase. Hopefully the base models will be retaining the prices longer. Memory/storage upgrades will definitely be higher.
 
  • Like
Reactions: mganu
Apple should buy some fabs and start making their own chips..

I have wondered that as well, for some time now. The future may see Apple buying a fab.

Apple is a 4 Trillion dollar company. You figure they would start making their own ram.

Maybe Apple should start manufacturing it's own RAM.

How would they do that? Please explain. All of the knowledge and tech needed to do this is tightly locked down by companies like TSMC. Even Apple doesn't know the process / recipe to make their chips. Making a CPU takes 3 months and requires literally hundreds of chemical, mechanical and baking steps that must be flawlessly executed.
This. I know Apple is very consumer product orientated these days, but that only applies to how they market themselves to the consumers, not the processes of actually making hardware..

Despite what far too many posters misassume, you can't just slap up a chipfab / foundry in the same way you can slap up a website on a server or a pop-up store in an empty retail space.

And you can't just hire a load of interns to fabricate chips. It's not casual / unskilled labor.

Semiconductors are not the same as Big Macs.
 
Last edited:
To put the increase in perspective for me, three years ago I upgraded from a 2012 to M2 mini. Also upgraded to two external Samsung T7 Shield SSDs. I was thinking of getting another one but seeing todays prices decided to offload some stuff onto an old HDD I have laying around.

2023
  • 2TB T7 Shield - $160
  • 4TB T7 Shield - $299

2026
  • 2TB T7 Shield - $490
  • 4TB T7 Shield - $1054
 
So with 90-100 Billion in profit each year Apple can't or won't build their own RAM factory? But Tesla/SpaceX can build a state of the art Fab Chip factory in Texas?
It will cost Tesla 20-25 Billion a year. It would be like the cost of several Apple Car projects worth per year! I don't think Apple wants to get into that, unless new leadership changes course from the past. They would sooner someone else build it or take over someone else's output etc.
 
More importantly, they are ruining the environment. Those new data centers won’t run on hope and happiness.

Trickle down energy shortages ahoy. Good old corporations. Passing the fun onto everyone else for a few quid in their coffers.

We’re fecked aren’t we?
Never mind all the farmland (aka food production) that will be permanently destroyed in the same process of building these new data centers! You start looking to all the issues, animals included, that are attached to these things and... it is just really bad for everyone.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.