Knowledge Vault 4 /70 - AI For Good 2022
Energy saving vs sustainable AI and data
Serge Conesa
< Resume Image >
Link to IA4Good VideoView Youtube Video

Concept Graph & Resume using Claude 3 Opus | Chat GPT4o | Llama 3:

graph LR classDef main fill:#f9f9f9, font-weight:bold, font-size:16px classDef Data_Centers fill:#d4e6f1, font-weight:bold, font-size:14px classDef Cooling fill:#d5f5e3, font-weight:bold, font-size:14px classDef Efficiency fill:#f9e79f, font-weight:bold, font-size:14px classDef Compatibility fill:#f5c6cb, font-weight:bold, font-size:14px classDef Space fill:#fadbd8, font-weight:bold, font-size:14px classDef Adoption fill:#f4d03f, font-weight:bold, font-size:14px classDef Longevity fill:#d4e6f1, font-weight:bold, font-size:14px classDef ROI fill:#d5f5e3, font-weight:bold, font-size:14px classDef E_Waste fill:#f9e79f, font-weight:bold, font-size:14px classDef Habits fill:#f5c6cb, font-weight:bold, font-size:14px classDef Decentralization fill:#fadbd8, font-weight:bold, font-size:14px classDef Energy_Savings fill:#f4d03f, font-weight:bold, font-size:14px classDef Data fill:#d4e6f1, font-weight:bold, font-size:14px classDef Solutions fill:#d5f5e3, font-weight:bold, font-size:14px classDef Sustainability fill:#f9e79f, font-weight:bold, font-size:14px classDef Urgency fill:#f5c6cb, font-weight:bold, font-size:14px classDef Direct_Cooling fill:#fadbd8, font-weight:bold, font-size:14px classDef Safe_Liquids fill:#f4d03f, font-weight:bold, font-size:14px classDef Waste_Heat fill:#d4e6f1, font-weight:bold, font-size:14px classDef Coordination fill:#d5f5e3, font-weight:bold, font-size:14px classDef Deployment fill:#f9e79f, font-weight:bold, font-size:14px A[Energy saving vs
sustainable AI and
data] A --> B[Data Centers:
Consumed 5% global
energy, 2016 1] B --> C[Immersion Cooling:
Efficient liquid
cooling for servers 2] B --> D[Air Cooling:
Inefficient, high
energy wastage 3] B --> E[Efficiency:
1500-3000x better
heat transfer 4] A --> F[Compatibility:
All servers,
needs SSDs 5] F --> G[Customization:
Optimal temperature
for components 6] F --> H[Elimination:
No raised floors,
fire suppression 7] F --> I[Space Savings:
70 m^2 for 500kW,
near-silent 8] A --> J[Adoption Issues:
Legacy designs,
resistance to change 9] J --> K[Repurpose:
Existing buildings
into data centers 10] J --> L[Longevity:
Extends server lifespan,
biodegradable liquids 11] A --> M[ROI:
Under 1 year,
reduces TCO 12] M --> N[E-Waste:
Addresses electronics
recycling crisis 13] M --> O[Habits:
Overcome bias,
proven technology 14] M --> P[Decentralization:
Enable edge computing,
renewables integration 15] A --> Q[Energy Savings:
Equivalent of
1500 1GW plants 16] Q --> R[Rethink Data:
Conservation, recycling,
efficient hardware 17] Q --> S[Ignored Solutions:
Actionable solutions
available today 18] A --> T[Sustainability:
Key criterion for
technology buyers 19] T --> U[Urgency:
Climate change, energy,
and water scarcity 20] A --> V[Direct Cooling:
Cool heat-generating
components in liquid 21] V --> W[Safe Liquids:
Non-toxic,
biodegradable, synthetic fluids 22] A --> X[Waste Heat:
Reuse for
other applications 23] X --> Y[Coordination:
Global priority for
sustainable infrastructure 24] A --> Z[Deployment:
Immediate action for
efficiency and sustainability 25] class A main class B,C,D,E Data_Centers class F,G,H,I Compatibility class J,K,L Adoption class M,N,O,P ROI class Q,R,S Energy_Savings class T,U Sustainability class V,W Direct_Cooling class X,Y Coordination class Z Deployment

Resume:

1.- Data centers consumed 5% of global energy in 2016, projected to increase to 20% by 2025, with high water usage and emissions.

2.- Immersion cooling technology can make data centers much more efficient by directly cooling servers in a liquid instead of air.

3.- Air cooling is very inefficient, requiring fans, air circulation, large facilities and complex cooling systems that waste a lot of energy.

4.- Liquid immersion cooling is 1500-3000x more efficient at heat transfer than air, reducing cooling energy needs by 95%.

5.- Any server can be made compatible with immersion cooling, only requiring a switch from hard drives to SSDs.

6.- Immersion cooling allows customizing the liquid temperature for each server component's optimal operating range, unlike uniform air cooling.

7.- It eliminates the need for raised floors, hot/cold aisles, fire suppression systems, and enables high density in small spaces.

8.- A 500kW air-cooled data center needs 270 m^2; the same capacity immersion-cooled only needs 70 m^2 and is near-silent.

9.- Lack of adoption is due to legacy air-cooled designs, inertia and resistance to change, despite immersion cooling's benefits.

10.- Immersion cooling can repurpose any existing building into a data center without extensive infrastructure changes.

11.- It extends server lifespans by avoiding temperature swings and vibrations, with synthetic liquids that are non-conductive and biodegradable.

12.- Return on investment is under 1 year for immersion cooling vs 10-15 years for air, with 30-90%+ reduction in TCO.

13.- Immersion can help address the growing e-waste crisis, where 70% of electronics go unrecycled and are often burned in developing countries.

14.- Widespread adoption requires shifting habits and overcoming status quo bias, but the technology is proven and ready to deploy at scale.

15.- Modular, decentralized immersion-cooled data centers can enable edge computing and better integration of variable renewable energy.

16.- The energy savings potential is massive - a wholesale shift could save the equivalent of 1500 1GW power plants globally.

17.- We need to rethink the growth of data from the ground up through conservation, recycling, algorithms, and efficient hardware.

18.- Policymakers and NGOs discuss long-term climate targets while ignoring actionable solutions available today like immersion cooling.

19.- Technology buyers need to make sustainability a key criterion and educate themselves on efficient options like immersion cooling.

20.- Inaction on data center efficiency is increasingly unacceptable given the urgency of climate change and scarcity of energy and water.

21.- Cooling servers is best done by cooling the heat-generating components directly in liquid, not by cooling the entire room air.

22.- Liquids used in immersion cooling are non-toxic, biodegradable, synthetic fluids that are safe and environmentally benign.

23.- Immersion cooling can make use of waste heat for other applications, further improving the efficiency of data infrastructure.

24.- We need World Bank-level coordination and long-term thinking to make sustainable digital infrastructure a global priority and standard.

25.- Solutions exist today to dramatically improve data center efficiency and sustainability, they just need to be deployed at scale immediately.

Knowledge Vault built byDavid Vivancos 2024