Memory, storage and photonic suddenly went ballistic and only a few saw them coming.
I am not someone working in the tech industry nor someone who knew ai revolution so well that I am able to even analyze and kinda foresee what might be in demand during this one of the most important time in human history. At a result, I would like to know what do you think might be coming for the next wave that hasn’t happened yet. What are components of infrastructure that you think might be in situation like memory and cpu where demand exceeds supply.
We all know that cpu and gpu are for inferences and training, photonics accelerate the process and then memory and storage to reduce latency, enhance computation and store data. And all these are needed for ai models and applications. And all these is what I know so far. All these are still in demand because computing capabilities are still in shortage, and for ai to work they are essentials, but what else?
What do you think might be the next components and essentials for ai revolution?
byu/VitoKan instocks
Posted by VitoKan
3 Comments
Anything energy. There literally isn’t not enough power production to support current planned / I construction data centers.
AI is currently turning into a full infrastructure buildout.
A few areas that could become the next bottlenecks:
Power and cooling: AI datacenters are running into energy limits fast
Networking / optical interconnects: moving data between GPUs is becoming critical. Think Verizon, Comcast
Data infrastructure: vector databases, synthetic data, retrieval systems
Edge AI chips: AI moving into phones, cars, robotics, devices
AI security: autonomous agents massively expand attack surfaces. AI enabled cybersecurity companies
I think the next wave is less about “who has the best model” (models are all going to get really really good) and more about who enables AI to scale reliably
Power and cooling infrastructure honestly