AI Forecast 2030 The Future Of AI
AI Forecast 2030 – The Future Of AI
Nvidia is finished
Soon the big tech will build their own chips and break free of nvidia. China will be the first to free itself from Nvidia. Nvidia will have 0% presence in China very soon.
Simultaneously startups will build better accelerators for inference and the entire world will shift to them. It takes only $20m to burn an accelerator today.
China is 1 generation behind the frontier models. It will release open source weights to better and better trillion parameter models at 10x cheaper costs killing OpenAi and Anthropic. All enterprises will shift to chinese models.
The future of AI is 100% Open Source. There will be nothing sovereign or propreitary or enterprise about it. Everyone will buy hardware and host it on prem or choose a pre-packaged one-click-deploy hosting service of their choice as per their convenience.
Small AI Models will be prevelant onboard across consumer devices very soon.
The next wave will be Open Hardware and a sprawl of 130nm kind of fabs everywhere. Even hobbyists will be able to fabricate their own chips in lot sizes of 10 or 100 pieces very very soon.
The progress of AI has stalled. MIT has proven that AI has nowhere to go from here. So every vendor in the world will essentially be peddling the same warez very soon. And open source will catchup and be on par very very soon and will be the ultimate winners of this race.
Cloud vendors fund startups and companies to exclusively adopt their cloud services. Nvidia funds billions of dollars to AI Vendors to buy its GPUs. they have relied on these circular arrangements to create a massive hype but such things suppress two things:
- there is no sustainable business demand for so many GPUs
- that AI is economically unviable
All these companies will crash and burn, oracle might be the first one to sink. Triggering a global recession.
Physical AI will flop in 3-4 years. Robotics will crash and burn too. Robots will be able to kill but unable to do anything meaningful reliably in the real world. They will do reasonably well in controlled environments like factories though.
Brain chip implants will never be able to write into the brain failing on its biggest promise of two way integration and augmented storage.
Intel is desperate for a comeback, and AMD wants it's share of AI both will get a good chance.
ARM will start losing out to RISC-V at an ever increasing pace.
Space Datacenters will be a flopshow. It is just a whole lot of marketing gas and hyperbole, the economics and logistics don't work out.
99% of agentic ai enterprise projects will fail to deliver quality, reliability, accuracy, customer experience and most of all economic viability.
SAAS will get exterminated and systems will be built inhouse onprem. Zoho, Salesforce will reduce by 100X or simply go extinct in 2 years.
Big tech will play the dirtiest games to use regulations, compliance and audits to kill any challengers aka Regulatory Capture.
US will upset and alienate it's entire first generation immigrants. 50% drop in students expected.
It might take an average Joe 3 years to get a job out of college.
IQ levels will drop 20% across the world in the coming years.
Deepfakes will wreck havoc.
Cyberattacks and cyberfrauds will increase 100X and will become unbearable and uncontrollable.
Indian and south east asian IT industry will die out. The world will have 1 million boutique it firms and corporates will mix and choose them per project requirements.
90% of AI services startups will go bankrupt and fail to make any profits in 10 years.
IPO will cease to be a viable exit as retail investors will mature and reject nonsensical businesses and their nonsensical valuations.
Acquihiring will make a major comeback.