Future of AI 🤯
Vložit
- čas přidán 27. 08. 2024
- 🌟 By 2030, AI workloads will transform! 🌟
🔹 Training will only account for 10% of compute demand, with 90% focused on inferencing.
🔹 AI data center demand will be 8x higher.
🔹 Over 2 billion AI PCs will be in use.
Jeff Clarke, COO at Dell, shared these insights at #DTW2024. Startups and enterprises are moving towards faster, cost-optimized AI models and hardware.
I’ve admired Vasudev Lal’s work in open-source generative AI. He predicts AI adoption in three stages:
1️⃣ Today: Age of AI Copilots - AI augments human capabilities.
2️⃣ Tomorrow: Age of AI Agents - AI automates domain-specific workflows.
3️⃣ Future: Age of AI Functions - AI automates entire enterprise departments. Intel's booth had an impressive demo using their Gaudi 2 processors to create superhero avatars.
The stats on their upcoming Gaudi 3 processors are exciting:
🚀 1.5x faster time-to-train Llama and GPT models vs. H100.
🚀 1.5x and 1.3x faster time-to-inference vs. H100 and H200.
Check out the full report: [link in bio]
#AI #TechInnovation #FutureOfWork #GenerativeAI #IntelGaudi3 #DTW2024 #IntelAmbassador
I agree with your prediction.
No waste of time. Excellent loved it❤
Nice info.
Regarding inference, does the algorithms will be improved, so inferences cost will be comparatively low.
And what may be some future techniques to reduce cost of inference.
Hello mam I want to ask you a question
Good afternoon mam
Disliking this video because I want to demotivate someone 😂