3 ms·

In the next 1-5 years we are going to see a lot of big companies falling apart. Small teams, with the help of tech/AI can now move so much faster than big team

by nico 4y ago

In the next 1-5 years we are going to see a lot of big companies falling apart. Small teams, with the help of tech/AI can now move so much faster than big teams. The communication and coordination overhead is starting to be too big of a burden, especially when you can have 1 person do the work of 30-100 (using AI).

danpalmer 4y ago
I’m not sure that’s true given hardware as a limiting factor. Small teams have always been able to move faster in building things, they have fewer requirements and less to lose. But small teams can’t buy expensive hardware up front, and modern AI is hardware constrained for many purposes. These companies will either end up renting hardware from cloud providers, or using AI APIs from cloud providers.
- nico 4y ago
  Sure, that’s an immediate limitation. Big hardware will go slower and come later. When AI has fully integrated with robotics.
garganzol 4y ago
Sarcastically, moving fast (while breaking things) is not the exact way to a success because nothing works at the end.
Glemotooo 4y ago
Hard to understand your comment in context of intel. There are a handful fabs around the whole world who can even make those chips intel and tsmc can do. No one will just replace any of them in any near capacity at all. And the entry to this level of tech is not 'just a few millions' ist actually (just making a fab with people and processies of people who know how to do this) is 10-20 billions. And if you would take any software company i also disagree: I was working in a small company, they didn't knew security really. I intriduced a ton of security. In comparision to the huge company i'm now: The developers still don't know much about security but the security teams of my company do. In my opinion the gap will be even bigger because small companies (besides the cool and hip new elite startups) still have more of a traditional mindset while the big onces need to compete. They need to use AI and co and they have the financial background to create AI teams.
davemp 4y ago
> when you can have 1 person do the work of 30-100 (using AI) That’s quite the statement. I highly doubt that will be reality in the next 1-5yrs, especially for novel R&D work. Even more so for products in the physical world.

by nico 4y ago

danpalmer 4y ago

I’m not sure that’s true given hardware as a limiting factor. Small teams have always been able to move faster in building things, they have fewer requirements and less to lose. But small teams can’t buy expensive hardware up front, and modern AI is hardware constrained for many purposes. These companies will either end up renting hardware from cloud providers, or using AI APIs from cloud providers.

nico 4y ago
Sure, that’s an immediate limitation. Big hardware will go slower and come later. When AI has fully integrated with robotics.

garganzol 4y ago

Sarcastically, moving fast (while breaking things) is not the exact way to a success because nothing works at the end.

Glemotooo 4y ago

Hard to understand your comment in context of intel. There are a handful fabs around the whole world who can even make those chips intel and tsmc can do. No one will just replace any of them in any near capacity at all. And the entry to this level of tech is not 'just a few millions' ist actually (just making a fab with people and processies of people who know how to do this) is 10-20 billions. And if you would take any software company i also disagree: I was working in a small company, they didn't knew security really. I intriduced a ton of security. In comparision to the huge company i'm now: The developers still don't know much about security but the security teams of my company do. In my opinion the gap will be even bigger because small companies (besides the cool and hip new elite startups) still have more of a traditional mindset while the big onces need to compete. They need to use AI and co and they have the financial background to create AI teams.

davemp 4y ago

> when you can have 1 person do the work of 30-100 (using AI) That’s quite the statement. I highly doubt that will be reality in the next 1-5yrs, especially for novel R&D work. Even more so for products in the physical world.