ENABLING ARTIFICIAL INTELLIGENCE WITH ENGINEERED SUBSTRATES
December 2024
"Artificial intelligence is a machine's ability to perform the cognitive functions we usually associate with human minds"
Source: McKinsey
|
1 AI FUNDAMENTALS |
2 3 |
© Soitec 2024. No copying or distribution permitted. |
PAGE5 |
ARTIFICIAL INTELLIGENCE IS TRANSFORMING OUR DAILY LIVES
|
Content |
Virtual |
Climate |
Smart Mobility |
|
Creation |
Assistants |
Research |
|
|
Offering new tools for |
Augmenting productivity |
Helping to combat |
Driving Automotive |
|
content creators |
in everyday tasks |
climate change |
Autonomy & Efficiency |
|
Healthcare & |
Wearables & |
Security & |
Industry |
|
Lifesciences |
Hearables |
Privacy |
4.0 |
|
Discovering new drugs & |
Delivering personal aid |
Enhancing threat |
Accelerating automation |
|
preventive treatments |
to overcome disabilities |
detection & prevention |
& efficiency roadmaps |
|
1 AI FUNDAMENTALS |
2 3 |
© Soitec 2024. No copying or distribution permitted. |
PAGE7 |
CHALLENGES FROM THE APPLICATIONS TO THE HARDWARE LAYERS
SOITEC TECHNOLOGY TO LEVERAGE CLOUD AI & EDGE AI NEW CHALLENGES
|
Architecture& |
Smart |
Image |
ADAS/AD |
SmartCity |
Voice |
Face |
Real-time sound |
Industry4.0 |
|
Infrastructure |
assistants |
generation |
recognition |
recognition |
processing |
|
Large Language Models |
Domain-Specific Models |
|
General purpose LLM and other models (such as |
AI models trained on specific data to perform tasks |
|
ChatGPT, DALL-E…) replicating human-like thinking |
with greater precision (enterprise, professional |
|
and decision-making processes |
content creation, simulated data, etc.) |
|
Hyperscale data centers and |
Edge AI chips running optimized AI |
|
|
enterprise servers powered by AI |
||
|
models at low power for lower |
||
|
accelerators running large models |
||
|
complexity tasks |
||
|
for highly complex tasks |
||
|
Cloud AI |
On-device |
|
|
infrastructure |
EdgeAI |
Source: Qualcomm, Red Hat
|
1 AI FUNDAMENTALS |
2 3 |
© Soitec 2024. No copying or distribution permitted. |
PAGE8 |
ARTIFICIAL INTELLIGENCE
EXPONENTIAL GROWTH IN COMPUTING POWER
AI ACCELERATION
|
$10-15T |
10x |
+80% |
|
POTENTIAL VALUE AT STAKE |
DEVICESRUNNING |
CHAT GPT USERS SINCEJAN'23 |
|
ARTIFICIAL INTELLIGENCE |
AI IN2030 |
3.7B WEBSITE |
|
~$10-15 TRILLION |
21B IN 2030 |
VISITORS IN OCT'24 |
|
vs 1.8B TODAY |
|
Healthcare diagnostic |
Autonomous driving |
Industry 4.0 |
Digital creation |
Source: McKinsey, Transforma Insights, OpenAI; Center for a New American Society, NamePepper
Acceleration of Computing Power
GPT-4 estimated computing power
|
power |
Hardware + Algorithmic |
||
|
1E+33 |
improvements |
||
|
Computingtraining (FLOPs) |
|||
|
1E+32 |
|||
|
1,000,000 x GPT-4 |
|||
|
1E+31 |
|||
|
1E+30 |
Hardware |
||
|
improvements only |
|||
|
1E+29 |
|||
|
ModelAI |
1E+28 |
1,000 x GPT-4 |
|
|
1E+27 |
|||
|
1E+26 |
|||
2022 2024 2026 2028 2030 2032 2034 2036 2038 2040 2042
|
1 AI FUNDAMENTALS |
2 3 |
© Soitec 2024. No copying or distribution permitted. |
PAGE10 |
| Attention: This is an excerpt of the original content. To continue reading it, access the original document here. |
Attachments
Disclaimer
Soitec SA published this content on December 04, 2024, and is solely responsible for the information contained herein. Distributed by Public, unedited and unaltered, on December 04, 2024 at 16:50:21.836.
