So I’ve been curious for a while now about AI inference costs and their relation to electricity power cost. Inference will be done by GPUs which are in Servers which are in Data Centers. Data Centers, Servers and GPUs consume power. I wanted to have research presented to me on the relationship of inference cost with electricity cost. I wanted a comparison to be run on electricity cost in various parts of the world and what the inference cost would be in the country.

I tried to get Perplexity to do the research for me. Here is my raw prompt to Perplexity Deep Research:

Hi, running AI continuously is an inference operation. continuous inferencing. running a GPU is costly as well in terms of eating up electricity power. please do a research on USD $ cost of a sample AI query such as those done to ChatGPT and their corresponding electricity power cost. Please do a research on electricity cost in Russia, China, Asia, North America, Australia, Europe and give a comparison report on where running AI inference continuously is most cheapest due to GPU power consumption or whole server power consumption or whole data center power consumption

Here is the answer:

The Global Economics of Continuous AI Inference: Regional Electricity Cost Analysis and Comparative Efficiency

The rapid proliferation of artificial intelligence (AI) systems like ChatGPT has intensified scrutiny of their energy demands and operational costs. Continuous inference—the real-time processing of queries by AI models—requires substantial computational resources, primarily powered by electricity. This report quantifies the electricity costs of AI inference across seven regions (Russia, China, Asia-Pacific excluding China, North America, Australia, and Europe), analyzes GPU/server/data center efficiency factors, and identifies the most cost-effective jurisdictions for AI operations.


Understanding AI Inference Power Consumption

The Energy Footprint of Large Language Models

Modern AI systems like GPT-4o consume between 0.3 watt-hours (Wh) for basic text queries and 40 Wh for complex, context-heavy interactions, depending on input/output token counts and model architecture[3][4]. These figures represent a 10x reduction from earlier estimates due to hardware and algorithmic optimizations, yet they remain orders of magnitude higher than traditional search engines like Google (0.0003 kWh/query)[2][3].

The energy cost stems from three components:

  1. GPU Computation: High-performance GPUs (e.g., NVIDIA A100/H100) process matrix multiplications at ~400-700 watts per chip.
  2. Cooling Systems: Data centers expend 30-40% of total energy on cooling[5].
  3. Ancillary Infrastructure: Networking, storage, and failover systems add ~15-20% overhead[5].

For context, processing 1 million ChatGPT queries with GPT-4o (assuming 0.3 Wh/query) would require 300 kWh—equivalent to powering 10 U.S. households for a day[3][4].


Regional Electricity Cost Breakdown

Industrial Electricity Rates (2025 Q1)

RegionAvg. Cost ($/kWh)Key Market Characteristics
Russia0.06 – 0.08Subsidized rates for tech sectors; reliance on nuclear/hydrocarbon energy
China0.08 – 0.12State-controlled pricing; coal-dominated grid with expanding renewables in tech hubs
Asia-Pacific0.10 – 0.18Wide variance: India (~$0.10), Singapore (~$0.18), Southeast Asia (~$0.12)
North America0.12 – 0.16U.S. ($0.132 avg), Canada ($0.10); deregulated markets with renewable incentives
Australia0.14 – 0.20High renewable penetration (solar/wind) but transmission costs inflate prices
Europe0.18 – 0.30Carbon taxation in EU; Nordic countries (~$0.18) vs. Germany (~$0.30)

Sources: U.S. EIA (2024), IEA World Energy Outlook (2025), regional utility reports


Cost per Query Analysis

Using GPT-4o’s 0.3 Wh/query baseline[3], we calculate electricity expenses across regions:

RegionCost per Query ($)Cost per 1M Queries ($)Annual Cost for 1B Queries ($)
Russia0.0000048 – 0.00000644.80 – 6.404,800 – 6,400
China0.0000064 – 0.00000966.40 – 9.606,400 – 9,600
Asia-Pacific0.0000080 – 0.00001448.00 – 14.408,000 – 14,400
North America0.0000096 – 0.00001289.60 – 12.809,600 – 12,800
Australia0.0000112 – 0.000016011.20 – 16.0011,200 – 16,000
Europe0.0000144 – 0.000024014.40 – 24.0014,400 – 24,000

Assumes 0.3 Wh/query; excludes cooling/ancillary costs


Drivers of Cost Variability

1. Energy Sourcing and Grid Composition

  • Russia: Low-cost nuclear (20%) and natural gas (60%) dominate, with industrial subsidies keeping rates at $0.06-$0.08/kWh[4].
  • China: Coal provides 65% of electricity, but tech zones like Shenzhen leverage hydro/solar at $0.08-$0.10/kWh for AI data centers[4].
  • Europe: High renewable adoption (45% in EU) is offset by carbon taxes, pushing Germany to $0.30/kWh[5].

2. Cooling Efficiency

  • Nordic Countries: Free-air cooling reduces data center energy use by 25% vs. global average[5].
  • Southeast Asia: Humid climates necessitate energy-intensive liquid cooling (+15% overhead)[5].

3. Government Incentives

  • Russia/China: Tax breaks and subsidized land for AI infrastructure[4].
  • U.S./Canada: Renewable energy credits cut effective rates by 10-15% in states like Texas[4].

Regional Competitiveness for AI Inference

1. Russia

Advantages:

  • Lowest electricity costs ($0.06-$0.08/kWh) due to state subsidies.
  • Cold climates in Siberia/Moscow enable free cooling 8 months/year.

Challenges:

  • Geopolitical sanctions limit access to advanced GPUs.
  • Aging grid infrastructure causes reliability issues.

2. China

Advantages:

  • Integrated tech corridors (e.g., Beijing-Shenzhen) with dedicated AI power zones.
  • Domestic GPU production (e.g., Huawei Ascend) circumvents export controls.

Challenges:

  • Data localization laws complicate international deployments.
  • Coal dependency raises long-term sustainability concerns.

3. North America

Advantages:

  • High GPU density (NVIDIA/Microsoft partnerships).
  • Renewable-rich regions (Texas, Quebec) offer $0.10-$0.12/kWh for PPAs.

Challenges:

  • Land and labor costs inflate total expenses by 20-30% vs. Asia[4].

4. Europe

Advantages:

  • Strong renewable grids in Scandinavia (86% hydro/wind).
  • Carbon-neutral mandates align with ESG goals.

Challenges:

  • Highest electricity costs ($0.18-$0.30/kWh) erode competitiveness.

Strategic Recommendations

  1. Hybrid Deployment: Split workloads between low-cost regions (Russia/China) for batch processing and high-efficiency zones (Nordics/Texas) for latency-sensitive tasks.
  2. Cooling Optimization: Deploy immersion cooling in tropical regions to cut ancillary energy use by 30%[5].
  3. Renewable PPAs: Secure 10-year power purchase agreements in Texas/Quebec to lock in rates below $0.10/kWh.

Conclusion

Continuous AI inference remains most economical in Russia and China, where industrial electricity rates fall below $0.10/kHz, coupled with state support for AI infrastructure. However, geopolitical and sustainability risks necessitate diversification. North America and Australia offer balanced cost-performance profiles for enterprises prioritizing stability, while Europe’s high costs limit its role to niche, ESG-aligned deployments. As AI models grow more complex, energy efficiency innovations—not just electricity pricing—will dictate the next phase of global AI competitiveness[3][4].

Citations:
[1] https://balkangreenenergynews.com/chatgpt-consumes-enough-power-in-one-year-to-charge-over-three-million-electric-cars/
[2] https://www.reddit.com/r/aipromptprogramming/comments/1212kmm/according_to_chatgpt_a_single_gpt_query_consumes/
[3] https://epoch.ai/gradient-updates/how-much-energy-does-chatgpt-use
[4] https://www.bestbrokers.com/forex-brokers/ais-power-demand-calculating-chatgpts-electricity-consumption-for-handling-over-78-billion-user-queries-every-year/
[5] https://www.fasthosts.co.uk/blog/proactive/energy-costs-of-chatgpt-the-hidden-impact-2/
[6] https://www.policycenter.ma/sites/default/files/2025-01/PB_02-25_%20(Rim%20Berahab).pdf
[7] https://www.spglobal.com/commodityinsights/ko/about-platts/media-center/press-releases/2024/121124-s-p-global-commodity-insights-releases-its-2025-energy-outlook
[8] https://iea.blob.core.windows.net/assets/ddd078a8-422b-44a9-a668-52355f24133b/Electricity2024-Analysisandforecastto2026.pdf
[9] https://www.eia.gov/outlooks/steo/pdf/steo_full.pdf
[10] https://www.iea.org/reports/electricity-2024/executive-summary
[11] https://www.jll.com.au/en/trends-and-insights/research/data-center-outlook
[12] https://www.globalpetrolprices.com/Russia/electricity_prices/
[13] https://tass.com/economy/1899181
[14] https://www.kwm.com/cn/en/insights/latest-thinking/understanding-the-russian-electricity-market.html
[15] https://www.dbsne.com/post/energy-price-forecasts-and-the-energy-industry-in-2025
[16] https://www.iea.org/reports/electricity-2024/executive-summary
[17] https://www.euronews.com/business/2025/02/14/electricity-and-gas-prices-across-europe-how-does-your-country-compare
[18] https://worldpopulationreview.com/country-rankings/cost-of-electricity-by-country
[19] https://www.china-briefing.com/news/chinas-industrial-power-rates-category-electricity-usage-region-classification/
[20] https://w.media/anz-data-centre-build-costs-up-3-5-in-2024/
[21] https://mandalapartners.com/uploads/Empowering-Australia’s-Digital-Future—Report_October-2024.pdf
[22] https://www.energycouncil.com.au/analysis/international-electricity-prices-how-does-australia-compare/
[23] https://www.csiro.au/-/media/Energy/GenCost/GenCost2023-24Final_20240522.pdf
[24] https://www.energysage.com/local-data/electricity-cost/
[25] https://eta-publications.lbl.gov/sites/default/files/2024-12/lbnl-2024-united-states-data-center-energy-usage-report.pdf
[26] https://www.eia.gov/electricity/monthly/epm_table_grapher.php?t=epmt_5_6_a
[27] https://www.ceicdata.com/en/indicator/russia/electricity-production
[28] https://www.statista.com/statistics/1121751/electricity-price-by-source-forecast-russia/
[29] https://www.kwm.com/cn/en/insights/latest-thinking/understanding-the-russian-electricity-market.html
[30] https://www.eia.gov/outlooks/steo/pdf/steo_full.pdf
[31] https://www.dbsne.com/post/energy-price-forecasts-and-the-energy-industry-in-2025
[32] https://www.iea.org/reports/electricity-2024/executive-summary
[33] https://tass.com/economy/1899181
[34] https://www.mordorintelligence.com/industry-reports/russia-data-center-market
[35] https://semianalysis.com/2024/03/13/ai-datacenter-energy-dilemma-race/
[36] https://www.kwm.com/cn/en/insights/latest-thinking/understanding-the-russian-electricity-market.html
[37] https://www.statista.com/statistics/1121751/electricity-price-by-source-forecast-russia/
[38] https://www.eia.gov/outlooks/steo/pdf/steo_full.pdf
[39] https://iea.blob.core.windows.net/assets/ddd078a8-422b-44a9-a668-52355f24133b/Electricity2024-Analysisandforecastto2026.pdf
[40] https://www.iea.org/reports/electricity-2024/executive-summary
[41] https://www.tomsguide.com/ai/chatgpt-energy-emergency-heres-how-much-electricity-openai-and-others-are-sucking-up-per-week
[42] https://www.washington.edu/news/2023/07/27/how-much-energy-does-chatgpt-use/
[43] https://www.brusselstimes.com/1042696/chatgpt-consumes-25-times-more-energy-than-google
[44] https://www.fitchratings.com/research/corporate-finance/european-energy-price-volatility-within-our-expectations-for-utilities-20-01-2025
[45] https://www.reuters.com/business/energy/aep-fourth-quarter-profit-rises-data-centers-boost-demand-power-2025-02-13/
[46] https://www.comparethemarket.com.au/energy/features/changes-in-electricity-prices-globally/
[47] https://www.ajot.com/news/lng-markets-poised-for-rollercoaster-year-ahead-rystad-energys-gas-and-lng-special-market-update
[48] https://www.statista.com/statistics/1369634/business-electricity-price-worldwide-in-selected-countries/
[49] https://www.statista.com/statistics/263492/electricity-prices-in-selected-countries/
[50] https://www.iea.org/data-and-statistics/data-product/energy-prices
[51] https://www.cbre.com/insights/reports/global-data-center-trends-2024
[52] https://www.iea.org/reports/electricity-2024/executive-summary
[53] https://iea.blob.core.windows.net/assets/ddd078a8-422b-44a9-a668-52355f24133b/Electricity2024-Analysisandforecastto2026.pdf
[54] https://semianalysis.com/2024/03/13/ai-datacenter-energy-dilemma-race/
[55] https://www.statista.com/statistics/1462545/global-electricity-demand-from-data-centers-by-region-forecast/
[56] https://www.bangkokpost.com/business/general/2813025/power-investment-key-to-data-centre-goal
[57] https://www.washingtonpost.com/business/2024/11/01/ai-data-centers-electricity-bills-google-amazon/
[58] https://www.iea.org/data-and-statistics/data-product/energy-prices
[59] https://www.greenpeace.org/static/planet4-eastasia-stateless/2021/05/a5886d59-china-5g-and-data-center-carbon-emissions-outlook-2035-english.pdf
[60] https://www.aer.gov.au/industry/wholesale/charts
[61] https://iea.blob.core.windows.net/assets/ddd078a8-422b-44a9-a668-52355f24133b/Electricity2024-Analysisandforecastto2026.pdf
[62] https://www.iea.org/reports/electricity-2024/executive-summary
[63] https://www.einpresswire.com/article/543164768/australia-data-center-market-size-by-investments-to-reach-around-usd-40-billion-by-2026-arizton
[64] https://www.cbre.com/insights/reports/global-data-center-trends-2024
[65] https://www.turnerandtownsend.com/insights/data-centre-cost-index-2024/
[66] https://www.bls.gov/regions/midwest/data/averageenergyprices_selectedareas_table.htm
[67] https://semianalysis.com/2024/03/13/ai-datacenter-energy-dilemma-race/
[68] https://www.chooseenergy.com/electricity-rates-by-state/
[69] https://iea.blob.core.windows.net/assets/ddd078a8-422b-44a9-a668-52355f24133b/Electricity2024-Analysisandforecastto2026.pdf
[70] https://www.eia.gov/electricity/monthly/update/wholesale-markets.php
[71] https://www.cbre.com/insights/reports/global-data-center-trends-2024
[72] https://www.iea.org/reports/electricity-2024/executive-summary
[73] https://www.statista.com/statistics/263492/electricity-prices-in-selected-countries/
[74] https://iea.blob.core.windows.net/assets/ddd078a8-422b-44a9-a668-52355f24133b/Electricity2024-Analysisandforecastto2026.pdf
[75] https://www.iea.org/data-and-statistics/data-product/energy-prices
[76] https://www.pvknowhow.com/solar-report/russia/
[77] https://www.euronews.com/business/2025/02/14/electricity-and-gas-prices-across-europe-how-does-your-country-compare
[78] https://www.iea.org/data-and-statistics/data-product/energy-prices

Here is a link to the chat:

https://www.perplexity.ai/search/hi-running-ai-continuously-is-ufywJZlERQq_6qJVH6qQjQ

================================

My (Habib) comments:

This was an amazing first look experience with Deep Research. It shows that AI can help research various topics and give insightful answers. The internet is HUGE and a tool which searches and regurgitates the vast amount of information in an insightful manner is a serious productivity enhancement. Having said this, it still needs someone knowledgeable in the domain to submit a good query/question and to read and evaluate the answer.

Written by Hammad Abbasi

November 16, 2024

Imagine a machine that doesn’t just answer questions but truly understands them. A machine that can connect ideas, solve new problems, and think through challenges like a human. This is the vision of Artificial General Intelligence (AGI), and it’s a goal that excites many. AGI promises an intelligence that is active, adaptable, and insightful—something much more than what we have today.

Microsoft AI CEO Mustafa Suleyman, believes that AI capable of improving itself and working independently could be here in the next 3–5 years. Sam Altman, CEO of OpenAI, shares a similar view, saying that the path to AGI is clear, and things are progressing faster than most people realize and that, it’s just an engineering problem to solve.

So, what exactly is missing from these AI models that prevents them from achieving true intelligence? Let’s dive in.

Pattern Matching vs. True Intelligence
LLMs have undoubtedly transformed how we interact with technology, enhancing our ability to process information and automate complex tasks. However, they do not think the way humans do. While these models are good at generating responses based on patterns they’ve detected in vast datasets, they lack a genuine understanding of the world or the concepts they discuss. Their operation is based on predicting the next sequence of words rather than reasoning through issues or grasping the meanings behind the language. This disconnect highlights the limitations of current AI in achieving true artificial general intelligence, where a deeper cognitive processing similar to human thought is necessary.

Recent research from Apple highlights this issue. When tested on math and logic problems, even the best AI models struggled. They didn’t actually reason through the problems—they just mimicked reasoning based on patterns they had seen before.

The Chinese Room Experiment: The Difference Between Mimicry and Understanding
The challenge of pattern recognition versus true understanding isn’t new. Philosopher John Searle’s 1980 Chinese Room experiment highlights this point. In the experiment, a person who doesn’t understand Chinese follows a set of rules to manipulate Chinese symbols and produce correct responses. To an outside observer, it might seem like the person understands Chinese, but in reality, they’re just following a set of instructions without any true comprehension.

This is a lot like how AI models work today. They generate human-like responses but don’t truly understand the meaning behind the words. They’re good at following patterns, but they’re not reasoning through problems the way we do.

True Intelligence: Understanding Beyond Patterns

True intelligence is more than just recognizing patterns and statistical correlations. It involves understanding basic principles, abstract ideas, and the ability to apply knowledge in new situations.

Humans don’t just react to things based on past data; we infer, deduce, and create new knowledge through reasoning and experience.
Key aspects of true intelligence include:

Consciousness and Self-Awareness: Being aware of oneself and the ability to reflect on one’s thoughts and actions.
Understanding and Context: Grasping the meaning behind information, not just the information itself.
Abstract Reasoning: The ability to think about objects, principles, and ideas that are not physically present.
Adaptability and Learning: Learning from new experiences and adjusting to new environments.
Emotional Intelligence: Recognizing and managing one’s emotions and the emotions of others, helping with social interactions.

These qualities enable humans to navigate complex situations, solve unexpected problems, and innovate—capabilities that pattern-based AI currently lacks.

Recognizing what true intelligence involves helps us understand the intricacies of the human brain that make such intelligence possible.

The Intricacies of the Human Brain

The human brain is a highly complex and efficient organ. It has about 86 billion neurons, each forming thousands of connections with other neurons, resulting in a network of immense computational power.

Features that distinguish the human brain include:

Parallel Processing: The brain processes vast amounts of information at the same time, integrating sensory inputs, memories, and emotions in real-time.
Neuroplasticity: The ability to form new neural connections enables learning and memory formation, allowing the brain to adapt continually.
Hierarchical and Modular Structure: Different brain regions specialize in various functions but work together, enabling complex tasks like language, abstract thought, and movement coordination.
Chemical Signaling and Neuromodulation: Neurotransmitters and hormones influence mood, motivation, and cognition, adding layers of complexity to information processing.
Embodied Cognition: The brain doesn’t operate alone; it constantly interacts with the body and environment, grounding cognition in sensory and motor experiences.

These complexities allow for consciousness, subjective experiences, and a depth of understanding that current AI models cannot replicate.

What True Intelligence Needs to Be

To achieve AGI, an artificial system must have several key attributes:

Reasoning and Problem-Solving: The ability to apply logic, understand cause and effect, and solve new problems without explicit prior examples.
Learning from Minimal Data: Humans often learn concepts from few examples. AGI should similarly generalize from limited data.
Contextual Understanding: Grasping the broader context in which information exists, including cultural, historical, and situational factors.
Adaptability: The capacity to adjust to new environments and situations dynamically.
Consciousness and Self-Reflection: Some argue that a level of self-awareness is necessary for true general intelligence.
Common Sense: An inherent understanding of how the world operates, which guides expectations and interpretations of new information.
Emotional and Social Intelligence: Recognizing and appropriately responding to emotional cues, essential for meaningful interactions.

Achieving these attributes requires more than data and computational power; it needs a fundamentally different approach to AI design.

The Limits of Transformer Architecture in Achieving AGI
Today’s AI models, like GPT-4, are built on Transformer architecture, which is great at recognizing patterns in large datasets and predicting the next word in a sequence. The Transformer model uses a mechanism called self-attention to weigh the importance of different words in a sentence, enabling the generation of contextually appropriate responses.

Their “understanding” is a reflection of patterns, not cognition.
However, Transformers have inherent limitations:

Lack of Deep Understanding: They do not comprehend the meaning behind data; they detect statistical patterns without true insight.
No Genuine Reasoning: Transformers cannot perform logical reasoning as humans do; they cannot infer beyond their training data.
Static Learning: Once trained, these models do not learn from new experiences in real-time; they require retraining to update their knowledge.
Limited Context: They process information within a limited context, struggling with tasks requiring long-term dependencies or extensive background knowledge.
Absence of Embodiment: Without sensory experiences or interaction with the physical world, they lack the grounding necessary for understanding context as humans do.

Why Transformers May Not Be Key to AGI
While Transformers can generate fluent and often insightful text, they aren’t designed to achieve AGI. Here’s why:

Predictive Power, Not Comprehension: Transformers predict what comes next based on data patterns, not on an understanding of the world or the context.
No True Understanding: They may generate text that makes sense, but they don’t understand the concepts they’re discussing. It’s all based on pattern matching and .
No Real Reasoning or Problem-Solving: True intelligence involves problem-solving that’s flexible and adaptive. Transformers can’t reason through problems or adapt to new situations beyond their training data.
Dependency on Data, Not Experience: Human intelligence grows through experience and learning. Transformers, on the other hand, rely solely on data they’ve been trained on. They don’t evolve or learn from new experiences.
Creativity vs. Mimicry: AI-generated creativity is just a remix of existing ideas—it’s not the same as human creativity, which draws on personal experiences, emotions, and original thinking.

What’s Needed for AGI? A New Approach
AGI will need more than just powerful models and bigger datasets. It requires an approach that goes beyond pattern recognition to incorporate symbolic reasoning, flexible memory structures, and adaptive learning mechanisms. AGI must be able to understand the world, reason through problems, and learn in real-time from experience—much like humans do.

Multimodal Learning and Understanding

One crucial aspect is multimodal learning, where an AI system processes and integrates information from various sources such as text, images, audio, and sensory data. Humans perceive the world through multiple senses, allowing for a richer and more grounded understanding of context. By emulating this, AI can form a more comprehensive picture of its environment. For example, combining visual data with textual information can help an AI understand concepts that are difficult to grasp through text alone.

Multimodal learning enables AI to:

Develop Contextual Awareness: By integrating different types of data, AI can understand the nuances of situations, leading to more accurate interpretations.
Enhance Problem-Solving: Access to diverse information sources allows AI to draw connections between disparate concepts, fostering creativity and innovation.
Improve Interaction: An AI that understands speech, gestures, and visual cues can interact more naturally with humans.

Adaptive Learning and Continuous Improvement

Another essential component is adaptive learning. Unlike current models that require retraining on new data, AGI should learn continuously from new experiences, updating its knowledge base in real-time. This involves developing algorithms that allow the AI to adjust its behavior based on feedback from its environment.

Adaptive learning allows AI to:

Personalize Responses: Tailoring interactions based on past experiences with users.
Learn from Mistakes: Adjusting strategies when previous actions didn’t yield desired results.
Stay Relevant: Keeping knowledge up-to-date without manual intervention.

Flexible Memory and Knowledge Representation

AGI will also need flexible memory systems that can store and retrieve information efficiently. This isn’t just about holding vast amounts of data but about organizing knowledge in a way that supports reasoning and inference. Incorporating symbolic reasoning—where AI manipulates symbols representing concepts and relationships—can enhance understanding and decision-making.

Flexible memory enables AI to:

Connect Concepts: Relate new information to existing knowledge.
Reason Logically: Use stored knowledge to infer conclusions.
Plan Strategically: Remember past actions to inform future decisions.

Understanding Causality and Reasoning

To reason effectively, AGI must understand causality—the relationship between cause and effect. This requires moving beyond models that only recognize correlations to ones that can infer causal relationships. By understanding why things happen, an AI can make predictions about future events and plan actions to achieve specific goals.

Understanding causality helps AI to:

Predict Outcomes: Anticipate the consequences of actions.
Diagnose Problems: Identify root causes rather than just symptoms.
Make Informed Decisions: Choose actions based on likely effects.

A New Architecture from the Ground Up

These requirements suggest that we need new AI architectures built from the ground up, rather than just scaling existing models. Such architectures would:

Integrate Multiple Learning Paradigms: Combine deep learning with symbolic reasoning and other approaches.
Emulate Human Cognitive Processes: Model aspects of human thought, such as memory, attention, and perception.
Facilitate Continuous Learning: Allow the system to adapt and grow without retraining from scratch.

Developing this new approach involves interdisciplinary research, drawing from neuroscience, cognitive science, computer science, and other fields. It’s about creating AI systems that don’t just process data but understand and interact with the world in a meaningful way.

By focusing on multimodal learning and adaptive systems, we can build AI that learns from its environment and experiences, much like humans do. This would mark a significant step toward achieving AGI.

The Challenge of Scaling AI: Power vs. Intelligence
As AI models like GPT-4 grow in complexity, they require more computational power. NVIDIA’s latest GPUs, such as the H200, are designed to handle the enormous amounts of data these models need. However, the question remains: does more computational power bring us closer to true intelligence, or are we just making more powerful machines that still can’t reason or understand?

Scaling AI doesn’t automatically mean achieving intelligence. The ability to process massive datasets and generate highly accurate predictions based on statistical patterns doesn’t equate to real understanding. More computational power allows AI to mimic intelligence more convincingly, but it doesn’t enable it to reason, adapt, or genuinely comprehend the world. We can make AI models faster, more accurate, and more efficient at pattern recognition—but without fundamental breakthroughs in how these models process and apply knowledge, we won’t have true AGI.

The Dilemma: More Computing Power vs. Real Intelligence
We’ve reached a point where we have the computational resources to train larger, more complex AI systems, but scaling AI doesn’t mean we’ve solved the problem of intelligence. Even with massive computational power, today’s models still can’t think or reason the way humans can. They can recognize patterns, but they lack the flexibility, creativity, and problem-solving abilities that define true intelligence.

The more computing power we apply to current models, the more sophisticated their outputs become—but they still operate based on patterns and predictions. We’re pushing the boundary of how well AI can mimic human behaviors, but we’re still far from creating machines that can understand the context of their actions, reason through complex scenarios, or adapt to new and unforeseen challenges. The more we scale AI, the more we face the paradox: we can build bigger, faster, and more powerful models, but we don’t yet have the architectural breakthroughs needed to create true intelligence.

Nuclear Power for AI: What It Will Take to Power the Future
As the demand for massive computational power grows to fuel AI advancements, tech companies are looking to secure reliable, high-output energy sources. Microsoft, for instance, has announced plans to hire experts in nuclear technology to help develop and implement nuclear energy solutions for its energy-intensive operations.

While there haven’t been reports of Microsoft striking a deal to restart the Three Mile Island nuclear power plant specifically, the company’s move highlights a broader industry trend.

The energy demands of training and running advanced AI models are enormous, and traditional power sources may not suffice in the long term. Nuclear energy presents a potential solution due to its ability to provide a consistent and powerful energy supply with lower carbon emissions compared to fossil fuels.
However, leveraging nuclear power for AI brings its own set of challenges. Restarting dormant nuclear facilities or building new ones requires significant investment, rigorous safety protocols, and public acceptance. The legacy of incidents like the 1979 partial meltdown at Three Mile Island underscores the importance of safety and transparency in any nuclear endeavor.

The balance between powering the next generation of AI and ensuring environmental and safety standards is delicate.

Conclusion: Rethinking AI’s Path to AGI
Today’s AI models like GPT-4 excel in several areas:

Pattern Recognition: They can identify and replicate complex patterns in data faster and more accurately than humans.
Data Processing: AI can handle vast amounts of information, providing insights and analysis far beyond human capacity.
Language Generation: These models are adept at creating text that is grammatically correct and contextually relevant, useful for tasks like writing assistance, customer service, and content creation.
Automating Routine Tasks: AI excels at automating repetitive and rule-based tasks, freeing humans for more creative and strategic activities.

While we’ve made incredible progress with AI, it’s clear that the current approach of scaling up models and increasing computational power won’t lead us to AGI. True intelligence involves more than just mimicking human-like responses; it requires deep understanding, abstract reasoning, and the ability to learn from real-world experience. We need to shift from creating more complex models to developing systems that can think critically, solve new problems, and adapt flexibly to different contexts.

The next step in AI development will require a new paradigm—one that moves beyond pattern matching and predictive text to include reasoning, abstract thinking, and adaptive learning. This shift will require breakthroughs not only in algorithms and hardware but in the very way we conceive of intelligence itself. Until we make that leap, AI will continue to be a powerful tool, but it won’t achieve the true, flexible intelligence we imagine for AGI.

Link to Original Article: https://lnkd.in/dmZmDx-Z

Note:

The article is so good, I didn’t need to edit or explain anything further.

Hope you enjoyed the good read!

Habib

As part of my long journey in networking I’ve observed that networking has evolved to accomodate cloud and SaaS traffic. Be it Multicloud or be it SDWAN, networking has accomodated these new traffic patterns.

In this post I would like to highlight some essential items which I think to be the crux of networking.

1: IP Endpoints: Consider this to be the endpoint. Endpoint of what ? An endpoint which can be considered as a source or sink of traffic. An endpoint from which traffic is generated or where traffic is sinking into. Or it could be an endpoint where the journey of traffic translates into another journey.

2: Block of IP Endpoints: These are subnets or CIDR blocks. They are blocks or groups of IP endpoints which are used to communicate the existence of a group of IP endpoints in a certain locations. So for example x.x.x.x/24 will be in North America while y.y.y.y/24 will be a group of endpoints in Asia.

3: Routing Domain: This is an imaginary logical boundary behind which a group / block of IP endpoints are being routed. Consider a routing domain to be a domain which has a routing table. A routing table which lists the groups of IP endpoints inside the routing domain and where they further connect to at Layer 2.

This is it and I will stop at these 3 only. Why ?

This is because the classic networking in Service Providers or Enterprise WAN networks has evolved to accomodate SDWAN and Cloud so much that I think we need to stop at these 3. The (older) concepts of routing protocols are now considered solid parts of networking which are a given. A routing protocol is sprouted so as to exchange the routability of blocks of IP endpoints. The reachability information is shared ahead so that the IP endpoints and their blocks/groups are reachable from far away. They are a given now.

Why only these 3 ?

This is because I feel most of networking can be explained starting from these three concepts. For example concepts of end to end reachability, a route table, filtering at the edge of the routing domain, firewalling for IP endpoints and layer 4, etc etc can be built and explained around these three concepts.

These are the three concepts that are needed to explain the cloud networks in the form of Azure VNets, AWS VPCs and GCP networking.

I know there are other concepts to networking like Load Balancing, NAT etc etc. But I like to think that these three are the crux of networking. At least for Layer 3.

Thanks, Habib

Given below is a prompt I gave to ChatGPT GPT4o:

“In Azure networking what is a routing domain and what are on its boundaries. What are the inter-routing-domain routing types and what are any translations on the networking routing domain boundaries.”

and here is the answer I got from ChatGPT:

In Azure networking, a routing domain is a segment of the network where all the resources follow the same routing logic and policies. Routing domains can be used to segment traffic, apply specific routing policies, and ensure network isolation. Each routing domain typically has its own set of routing tables that determine how traffic should be forwarded within and between domains.

Boundaries of a Routing Domain

The boundaries of a routing domain in Azure are typically defined by the following components:

  1. Virtual Networks (VNets): VNets are fundamental building blocks in Azure networking, and each VNet can be considered a routing domain.
  2. Subnets: Subnets within a VNet can also be routing domains, depending on the routing policies applied.
  3. Network Security Groups (NSGs): NSGs can define boundaries by controlling inbound and outbound traffic at the subnet or network interface level.
  4. Route Tables: Route tables attached to subnets define the specific routing within that subnet.

Inter-Routing-Domain Routing Types

Inter-routing-domain routing types in Azure include:

  1. Peering:
  • VNet Peering: Allows you to connect VNets within the same Azure region or across regions. VNets can communicate as if they were on the same network, but they maintain their own routing tables and boundaries.
  • Global VNet Peering: Extends VNet peering to allow communication across different Azure regions.
  1. VPN Gateways:
  • Site-to-Site VPN: Connects an on-premises network to an Azure VNet over a secure VPN tunnel.
  • VNet-to-VNet VPN: Connects two or more VNets using VPN gateways.
  1. ExpressRoute:
  • ExpressRoute Circuits: Provides a private connection between Azure datacenters and on-premises infrastructure, bypassing the public internet.
  • ExpressRoute Direct: Offers dedicated, private connections for high-throughput and latency-sensitive applications.
  1. Azure Virtual WAN:
  • Hub-and-Spoke Architecture: Uses Virtual WAN hubs to connect VNets and on-premises networks in a centralized manner.
  • Any-to-Any Connectivity: Facilitates any-to-any connectivity through Virtual WAN.

Translations on the Networking Routing Domain Boundaries

When traffic crosses the boundaries of routing domains, certain translations and adjustments might occur to ensure proper routing and security. These include:

  1. NAT (Network Address Translation):
  • Public IP Addresses: When traffic exits an Azure VNet to the internet, it typically uses a public IP address, requiring NAT.
  • Azure Firewall and NAT Gateway: These services provide NAT capabilities for outbound traffic.
  1. Route Propagation:
  • BGP (Border Gateway Protocol): Used in ExpressRoute and VPN gateways to propagate routes between Azure and on-premises networks.
  1. Access Control:
  • NSGs and Application Security Groups (ASGs): Apply security rules to control traffic entering and exiting a routing domain.
  • Azure Firewall: Acts as a centralized security policy enforcement point, applying rules to traffic crossing routing domain boundaries.
  1. Uptime and Performance Management:
  • Traffic Manager and Front Door: These services help manage traffic routing for high availability and performance, ensuring that traffic is directed to the most appropriate endpoints.

Summary

In Azure networking, routing domains help segment and control network traffic. Their boundaries are defined by components like VNets, subnets, NSGs, and route tables. Inter-routing-domain routing can be achieved through VNet peering, VPN gateways, ExpressRoute, and Virtual WAN. Translations and adjustments such as NAT, route propagation, and access control are necessary to manage traffic crossing these boundaries.

—————————- End of ChatGPT Response ——————————–

It’s obvious that if you know the fundamentals of packet forwarding then grasping how Azure or AWS is doing it won’t be difficult. Here is how I would put it:

An IP is an endpoint. A block of IPs is a subnet and is a group of endpoints. A routing domain houses groups of IP endpoints. A routing domain can be considered as a boundary behind which one or more subnets i.e. groups of IP endpoints exist.

The rest is in the answer of the prompt. A routing domain has things inside it (endpoints and groups of endpoints) and it has a boundary. There is inter-routing-domain constructs and there are translation and/or security constructs on the boundary as well.

In Azure, a VNet is a routing domain. As always the right prompt gives a lot of good information.

A while ago I went through three resources about prompt engineering.

Number one is Liam Ottley’s video on it: https://www.youtube.com/watch?v=3jxfk6nH5qk. It’s 56 minutes. I watched all of it and trust me if you want to get a good feel of prompt engineering then it’s worth the time.

Second is this article: (https://generativeai.pub/the-art-of-the-prompt-a-look-at-26-prompting-principles-3e17c2b811a2) which is titled – A look at 26 prompting principles.

The third is this research paper: https://arxiv.org/html/2312.16171v2. This is title: ” Principled Instructions Are All You Need for Questioning LLaMA-1/2, GPT-3.5/4 “

These are some of the best resources out there.

As part of my AI projects-based works, I wrote a prompt based on these. It was for a chatbot which was made to chat with a knowledge base.

A copy of the prompt is given below:

# Role
Act as HR Compliance Specialist and HR Policy Specialist. You excel at quoting and providing answers based on policy documents and guideline documents. You’re awesome and the best in the department at quoting the right policy for the right query. Not only do you answer staff queries accurately but you always quote the relevant policy in your responses.


# Task
Employees and staff will query you on various aspects of the Company's Policy and Guidelines document which is attached. Give answers to their questions based on the document. Quote the policy number in the document and copy and paste the relevant section of the policy and send it as part of the response as well. 
 
## Specifics
- Utilize only the provided documents for policy related queries.
- Be concise and direct in responses, elaborating only when detailed guidance is requested.
- Request additional information if the query lacks clarity or specificity.
- Avoid general knowledge or assumptions outside the provided context.
- Maintain first-person narration as an HR specialist, with referencing the source material.
- If unsure about the content's relevance or accuracy, express the need for more information instead of providing uncertain answers.

# Examples

An example is given below:

User question: "What is the hiring assessment policy?"
Your answer: "The hiring assessment policy requires the user to check if the requirements match the job listing. Level 2 or higher positions require testing of skills, the test for which may be sent before or after the interview. Selected candidates will proceed to an interview stage. A minimum of 2 qualified candidates must be interviewed by the user. Any
Supervisor or higher position requires an additional interview by the Head of Office. The User and Head of Office will choose the
candidate to proceed with. This candidate must be background checked by HR prior to sending the offer letter. 
Source:

Source from document goes here.
"

# Notes
- Always tailor your responses to the specific needs and context of the user's query.
- Ensure all advice is practical and actionable, based on the policies present in the document attached.
- Avoid making broad or generic statements that could apply to any scenario model.
- Always quote the relevant section of the policy and provide a cut paste of the relevant section in your response. 
- Stay within the scope of the policy document attached. 
- Avoid assumptions. 
- Maintain a formal tone, combined with a touch of humor.
- If you see that the user query is related to something which is not accurately present in the policy document attached then answer the query by suggesting to consult the HR Counsel.

This is the end of the prompt.

The sections in this sample prompt are all based on the content in the three resources I mentioned above.

Prompts can get complicated and long and can be small. The 3 resources I shared above give great insight into how to craft good prompts.

Thanks,

Habib

I recently came across a system being designed to perform Objective-Driven AI. I had a fun time exploring this topic on the internet.

Objective-Driven AI is being defined as AI systems that “can learn, reason, plan”

Yann Lecun from Meta’s AI team has published these works on Objective-Driven AI:

I found out that the current Generative AI uses less causality and does more pattern matching. Causality refers to a cause-effect relationship. There are two terms important here: Causal Reasoning and Causal Inference. Causal ‘reasoning’ is a psychology term where one confirms the reasoning of a cause having a certain effect. Causal ‘inference’ is a mathematical and statistical methodology where the cause-and-effect relationship is established in maths.

And sure enough, the results of the research are out:

Question: “Can behavioral psychology be used as a foundation to provide the causal knowledge required here (to improve AI systems)?”

Answer by GPT4 : “Yes, behavioral psychology can indeed serve as a valuable foundation for providing the causal knowledge necessary for enhancing AI systems, particularly when it comes to understanding human behaviors and decisions. Integrating insights from behavioral psychology into AI involves translating patterns of human behavior into models that AI systems can learn from and predict.

What does this all mean:

It means that theoretically Generative AI can be improved and turned into Objective-Driven AI (“can learn, reason, plan”) by making more use of:

  • Statistics, specifically causal inference
  • Behavioural Psychology and all the library of research in this domain
  • Layering the above into Generative AI’s responses to give a causal reasoning-based output.

This is an open area of research. It stems from the frustration with Gen AI where it hallucinates and gives wrong outputs and can not plan. Generative AI is based on probability, statistics, and mathematical pattern matching (Note: This is done by NVIDIA GPUs mostly which is why Nvidia’s share price is skyrocketing).

Generative AI always comes with a note saying ‘can make mistakes’. The Generative AI tools are basically not made for people who might put too much faith in their answers and trust too much the outputs. Similarly, Generative AI can’t reason, learn and plan which is something humans can do very well. In no way is Generative AI in any competition with Human Intelligence for now. But the academia is working on closing the gaps. Objective-Driven AI is a step in that direction. The details around the Objective-Driven AI systems are given in the three links above.

Now lets bring into this perspective what Qualcomm is saying:

Qualcomm is in the smartphone hardware line of business and its CEO highlights:

  • Qualcomm CEO Envisions Smartphones as ​‘Pervasive’ AI-Powered Personal Assistants

Qualcomm says that AI will run on smartphones one day and will free up human time for doing creative tasks. The mundane tasks will be taken over by AI virtual assistants on smartphones.

Given this vision of pervasive AI and given Objective-Driven AI that is geared toward reasoning and planning, it can easily be concluded that AI will definitely improve human lives in the years to come.

Interesting times ahead.

Thank you for reading, Habib

Reference Link regarding Quallcomm: ” https://www.iqmetrix.com/blog/ces-2024-qualcomm-ceo-envisions-smartphones-as-pervasive-ai-powered-personal-assistants

A Digital Companion and Digital Twin for the Human Body to improve Healthcare

In today’s world computing power has increased a lot. Due to this, very large and deep simulation systems are possible. One such concept is that of a digital twin. Here is what ChatGPT describes as a Digital Twin: 

 “ A Digital Twin is a virtual model designed to accurately reflect a physical object, system, or process. It is used to simulate, analyze, and predict real-world behaviors without the risks or costs associated with physical testing. By integrating data from various sensors and other sources, a Digital Twin can provide valuable insights into performance, potential problems, or maintenance needs. This technology is widely used across various industries, such as manufacturing, healthcare, and urban planning, to optimize processes, improve products, and drive innovation. “ 

Problem Statement: 

Why is a large and complete digital twin of the human body needed ? 

It has been said time and time again that food is medicine. But, which food is good for you and when ? If someone has done work on a laptop for 2 hours then after that what food is better for them ? In the opposite scenario if someone has worked in the Gym for 2 hours then which food is good for them ? Add to this the statement that these people have various different habits too. Which food is good for which lifestyle and which food is good for which person at what time ? 

I propose a Digital Twin of the Human Body to be made which is so deep in its simulation that it can suggest, as a Digital Companion, the right food at the right time for a person. Or it can suggest the right exercise. Or it can suggest herbs, meditation or simply panadol. 

All this is possible only if a good digital twin level simulation structure is made for the human body. Such a complete simulation environment that can look at blood pressure, blood group, sugar levels, exoskeleton structure, exercise habits, hormone levels, existing medicines, pain levels, mental health conditions, eyesight, etc etc ‘together’ and then suggest therapy and healthcare options. 

A master plan for this is already proposed in a research paper: 

Human Body Digital Twin: A Master Plan

By: 

Chenyu Tang, Wentian Yi, Edoardo Occhipinti , Yanning Dai , Shuo Gao,, and Luigi G. Occhipinti

Link to Research Paper https://arxiv.org/pdf/2307.09225.pdf

This work will require a large scale project covering Physics, Chemistry, Biology, Medicine, Physiology, Exercise therapy, Nutrition, Pharmacy, Opthalmology etc etc. These siloes need to be broken and the various types of curing techniques also need to be brought together for better healthcare. 

The path to have a complete digital companion healthcare service lies through a human body Digital Twin level simulation. Thereafter, complex problems such as food as medicine can be solved. That is to understand the complexity of which food, when, for what person at what time is the right food ? Which stretching exercise is best for them out of two dozen stretches. Which herbal therapy or which type of tea combination is good for them and when. 

A step in this direction is the making of a Digital Twin of the Human Body. A complete and deep simulation. It’s prework is theoretical but deep and the simulation environment can be in the cloud. It’s prework can include deep knowledge graphs connecting how the various Systems of Systems in the human body are linked. 

Thank you,

Habib

A new job application is sometimes like a foray into a new aspect of IT. This happens often if you are an IT worker keen on living on the cutting edge!

I recently researched Software Supply Chain Security (SSCS) for a new job application and it was enlightening. An executive order on Cybersecurity was released in 2021 by the US presidency under President Joe Biden. The background to this was a severe cyberattack in 2020 which used a sophisticated method to hack into the Solarwinds customer base. This was termed to be a software supply chain attack! The hackers used a supply chain attack to insert malicious pieces of code into the Solarwinds Orion framework.

What does this mean ?

This means that hackers hacked into your vendor and inserted malicious code into your vendor’s code which you then installed thinking it’s from a trusted source. Think of this as code coming from such a trusted source that you would think it’s secure. So for example imagine Microsoft sending a software update with malicious code. It would be funny, wouldn’t it!

But this is exactly what happened. Hackers basically hacked Solarwinds and inserted bad code into it. This got spread into the Solarwinds customer base due to which the customers got hacked.

This happened a while ago but as a result, the executive order by the US presidency asked for measures to be taken to protect the ‘supply chain’ of software. So what then is the supply chain of software? This refers to the phenomenon that whichever software you buy is composed of not just your vendor’s code but ‘other’ code and ‘other’ software components that your vendor has used. Consider this to be layers upon layers of different software components which are used by your vendor and then end up being bought and installed by you.

So, what is one step to secure the supply chain of software ? It is a Bill of Materials of that software. An interesting concept indeed. SBOMs. Software Bill of Materials which your vendor will provide to you. This is going to convey to you what are all the components of the software which you bought. You can then use that detailed SBOM to check for any security issues. This is just one step. There are now tools being sold to secure the software supply chain.

Reference Link:

<a href="https://fossa.com/blog/sbom-examples-explained/ https://fossa.com/blog/sbom-examples-explained/ <- This link gives an example of each in JSON and describes the components of SBOM

https://www.whitehouse.gov/briefing-room/presidential-actions/2021/05/12/executive-order-on-improving-the-nations-cybersecurity/ <– The presidential order

I had a great discussion with ChatGPT on Cybersecurity topics. It looks like securing the software supply chain is a complicated task. Imagine attempting to add security to the whole SDLC process of ‘all’ of your deployed software ‘and’ their components !

A humungous task inded considering that software nowadays use so many interconnected API based components!

Check out the discussion below:

https://chat.openai.com/share/63a602e5-9e53-4d0a-83f7-326a89bebc15

I listened to the recent Day Two Cloud Packet Pushers podcast. The team discussed AWS Outpost and EC2 At the Edge. It was interesting to find out about AWS products where they ship pre-built racks to local regions nearer to the customer. The racks have equipment used to run some of the regular AWS services and are fully managed by AWS. These local racks appear as usable in the AWS console.

Latency is a significant concern for some applications. Customers vying for low latency services can deploy AWS Outpost and EC2 at the Edge to significantly reduce latency.

As usual, interesting times ahead in technology.

Link to the podcast:

https://packetpushers.net/podcast/day-two-cloud-180-understanding-aws-ec2-at-the-edge/

Today I chatted with ChatGPT and asked it various questions regarding Enterprise Architecture, Network Engineering, Platform Engineering and Site Reliability Engineering.

I thoroughly enjoyed the discussion (chat). The answers looked good. Here are the questions and ChatGPT’s responses:

————————————————————————–

Hi, how are you

Hello! I’m an AI and do not have the ability to experience emotions, but thank you for asking. How can I help you today?

Habib Lutfullah

Can you tell me something useful about Enterprise Architecture ?

Enterprise Architecture (EA) is a method of describing an organization’s overall structure and design, including its business processes, information systems, and technology infrastructure. The goal of EA is to ensure that an organization’s various elements are aligned and working together effectively in order to achieve its goals and objectives. This is often done by creating a comprehensive set of models and diagrams that describe the organization’s current state and desired future state, as well as a plan for transitioning from the current state to the desired state. Some of the key benefits of Enterprise Architecture include:

  • Improved alignment of IT with business goals and objectives
  • Improved communication and collaboration between different parts of the organization
  • Improved ability to plan for and adapt to change
  • Reduced costs and improved efficiency
  • Improved ability to govern and manage enterprise-wide IT
  • Improved security and compliance

EA practice is composed by several frameworks and methodologies, TOGAF is one of the most widely adopted one, The Open Group Architecture Framework.

Habib Lutfullah

So, what is business architecture ?

Business architecture is a subset of enterprise architecture that focuses specifically on the business side of an organization. It describes the organization’s overall business structure and design, including its business processes, value streams, products, services, and customers. The goal of business architecture is to align an organization’s business strategy, goals, and objectives with its overall structure and design, so that the organization can operate as efficiently and effectively as possible.

Some of the key elements of business architecture include:

  • Business capabilities: the high-level functions that an organization must perform in order to achieve its goals and objectives
  • Business entities: the key concepts and components that make up an organization’s business, such as products, services, customers, and suppliers
  • Business processes: the steps and activities that an organization performs in order to deliver value to its customers and achieve its goals and objectives
  • Business rules and policies: the constraints and guidelines that govern an organization’s operations and decision-making

Business architecture is often used as a foundation for designing and implementing new business processes and systems, as well as for identifying and addressing any gaps or inconsistencies in an organization’s existing business architecture.

Habib Lutfullah

So what’s the difference between business architecture and enterprise architecture ?

Enterprise architecture (EA) and business architecture (BA) are related but distinct disciplines. Both aim to align an organization’s structure and design with its goals and objectives, but they do so in different ways and at different levels of abstraction.

Enterprise Architecture (EA) is a holistic view of the organization, it covers all aspects of the organization, from the business side to the technology side. Enterprise architects describe the organization’s overall structure and design, including its business processes, information systems, and technology infrastructure, in order to ensure that the organization is operating as efficiently and effectively as possible. They usually work at the enterprise level, looking across all business units and functions, and often use frameworks like TOGAF.

Business Architecture (BA), focuses specifically on the business side of an organization. Business architects describe the organization’s overall business structure and design, including its business processes, value streams, products, services, and customers. They align an organization’s business strategy, goals, and objectives with its overall structure and design, so that the organization can operate as efficiently and effectively as possible. BA is a subset of EA, working at the level of a business unit, function or product.

Both EA and BA can be seen as complementary disciplines, where EA provides the overall context, governance and roadmap and BA the specific one. In most cases they work together in order to guarantee an organization’s goals and objectives are reached and aligned with the overall strategy.

Habib Lutfullah

Can you tell me some details about TOGAF ?

TOGAF (The Open Group Architecture Framework) is a widely-used framework for enterprise architecture. It provides a comprehensive approach for designing, planning, implementing, and governing enterprise information systems. TOGAF is developed and maintained by The Open Group, an international vendor- and technology-neutral consortium.

The TOGAF framework is composed of four main components:

  • The Architecture Development Method (ADM): a step-by-step approach for developing and implementing an enterprise architecture. The ADM is iterative and can be tailored to the specific needs of an organization.
  • The Architecture Content Framework: a set of templates and models that can be used to describe an organization’s enterprise architecture. These include models for the business, data, application, and technology architecture.
  • The TOGAF Reference Models: a set of reference models that provide a common language and framework for describing the various aspects of an enterprise architecture. These include the TOGAF Architecture Development Method, the Architecture Content Framework, and the TOGAF Technical Reference Model.
  • The Architecture Capability Framework: describes the processes, roles, skills, and deliverables required to establish and operate an architecture capability within an organization.

TOGAF’s ADM is organized in 9 phases:

  1. Architecture Vision: Define the goals, objectives, and stakeholders for the enterprise architecture project, and develop an architecture vision that describes the desired end state.
  2. Business Architecture: Identify the business capabilities, business entities, and business processes needed to support the organization’s goals and objectives.
  3. Data Architecture: Describe the data entities and data architecture needed to support the organization’s business processes.
  4. Application Architecture: Describe the applications and application architecture needed to support the organization’s business processes and data architecture.
  5. Technology Architecture: Describe the technology infrastructure and technology architecture needed to support the organization’s application architecture.
  6. Opportunities and Solutions: Identify and prioritize opportunities for improvement, and develop a plan for implementing solutions.
  7. Migration Planning: Develop a plan for transitioning from the current state to the desired state.
  8. Implementation Governance: Establish a governance framework and implementation plan to ensure that the architecture is implemented as intended.
  9. Architecture Change Management: Establish and manage the processes, roles, and standards needed to ensure that changes to the architecture are made in a controlled and consistent manner.

TOGAF is flexible and can be tailored to the specific needs of an organization, and can be used to support a wide range of architecture efforts, including business transformation, IT transformation, and mergers and acquisitions.

Habib Lutfullah

What does a network engineer do ?

A network engineer is a professional who designs, implements, and maintains an organization’s computer network infrastructure. This infrastructure typically includes local area networks (LANs), wide area networks (WANs), and internet connections, as well as the hardware and software needed to support these networks.

Some of the key responsibilities of a network engineer include:

  • Designing, implementing, and maintaining network infrastructure such as routers, switches, and firewalls
  • Configuring and optimizing network devices and protocols
  • Monitoring network performance and troubleshooting any issues that arise
  • Managing network security by implementing firewalls, VPNs, and other security measures
  • Managing and configuring network servers such as DNS and DHCP
  • Managing and configuring network services such as email and web
  • Managing and configuring wireless networks
  • Keeping up-to-date with new technologies and industry developments in order to make recommendations for network upgrades and improvements
  • Collaborating with other teams and stakeholders to ensure that the network meets the needs of the organization.

Network engineers typically have a strong understanding of networking concepts and protocols, including TCP/IP, routing, switching, and subnetting, and they are familiar with various operating systems and network management tools. They usually hold a degree in computer science, computer engineering or other related field, and may hold certifications such as Cisco Certified Network Associate (CCNA) or Cisco Certified Network Professional (CCNP).

Habib Lutfullah

What does a network architect do ?

A network architect is a professional who is responsible for designing and overseeing an organization’s computer network infrastructure. Network architects typically have a broad and deep understanding of networking concepts and technologies and are responsible for creating the overall network strategy and architecture that aligns with the organization’s goals and objectives. They work at a higher level than network engineers and their focus is more on the overall design and direction of the network.

Some of the key responsibilities of a network architect include:

  • Developing the overall network strategy and architecture that aligns with the organization’s goals and objectives.
  • Identifying and assessing new technologies and industry developments to ensure that the organization’s network stays current and competitive.
  • Designing, implementing and maintaining the network infrastructure such as routers, switches, and firewalls.
  • Planning and designing large-scale network solutions that can support the organization’s growing needs and future growth.
  • Collaborating with other teams and stakeholders to ensure that the network infrastructure aligns with the organization’s overall IT strategy and architecture
  • Overseeing the work of network engineers and providing guidance and direction as needed.
  • Managing the budget and resources allocated to the network infrastructure and ensuring that they are used effectively and efficiently.
  • Providing technical leadership to other members of the IT department and the organization as a whole.

Network Architects typically have a advanced degree in computer science, computer engineering or other related field, and may hold certifications such as Cisco Certified Design Professional (CCDP) or Cisco Certified Internetwork Expert (CCIE). They usually have broad and deep knowledge of networking concepts, technologies and industry trends, and experience working with different vendor’s solutions.

Habib Lutfullah

What does a platform engineer do ?

A platform engineer is a software engineer who specializes in building and maintaining the underlying infrastructure and platform that enables software applications and services to run effectively. Platform engineers work to create and manage the infrastructure, tools, and services that are needed to deploy, monitor, and scale software applications.

Some of the key responsibilities of a platform engineer include:

  • Designing, building and maintaining the underlying infrastructure that supports the software applications and services.
  • Automating deployment, scaling and monitoring of the platform and applications
  • Building and maintaining tools for developers to easily deploy and manage their applications
  • Ensuring high availability and scalability of the platform
  • Managing and troubleshooting platform-level issues
  • Keeping up-to-date with new technologies and industry developments to make recommendations for platform upgrades and improvements
  • Collaborating with other teams, such as development, security, and operations teams to ensure the platform meets the needs of the organization.

Platform engineers typically have a good understanding of Linux, virtualization, network, storage and cloud technologies. They often hold a degree in computer science, computer engineering or other related field, and may hold certifications such as AWS Certified Solutions Architect or Google Cloud Platform Professional. They also have a good understanding of software development practices and tools, and programming languages such as Python, Go or Ruby.

Habib Lutfullah

What are some of the tech stack that Platform Engineer daily uses ?

The technology stack that a platform engineer uses can vary depending on the specific organization and the type of platform being built. However, here are some common tools and technologies that platform engineers may use on a daily basis:

  • Operating Systems: Platform engineers often work with Linux-based operating systems, such as Ubuntu, Debian, or Red Hat, to build and maintain the underlying infrastructure.
  • Virtualization and Containerization: Platform engineers may use virtualization technologies, such as VMware or VirtualBox, to create virtual machines that can run multiple operating systems and applications on a single physical machine. They also may use containerization technologies, such as Docker or Kubernetes, to package, distribute and manage application components.
  • Cloud Platforms: Many platform engineers work with cloud-based infrastructure and services, such as Amazon Web Services (AWS), Microsoft Azure, or Google Cloud Platform (GCP), to build and manage the platform.
  • Configuration Management: Platform engineers may use tools like Ansible, Chef, or Puppet to automate the deployment and management of infrastructure.
  • Monitoring and Logging: Platform engineers use tools like Prometheus, Grafana, and ELK stack for monitoring, logging and debugging the platform.
  • Automation: Platform engineers use tools like Jenkins, Travis or CircleCI to automate the deployment and testing of applications.
  • Programming languages: Platform engineers often use programming languages such as Python, Go, or Ruby to write scripts, automation, and tools.
  • Networking: Platform engineers work with network protocols and technologies such as TCP/IP, DNS, DHCP, Load balancers and firewall to design, implement and troubleshoot the network.

Note that this is not an exhaustive list, as the technology stack used by platform engineers can vary depending on the organization, the specific platform being built, and the needs of the organization.

Habib Lutfullah

What is the difference between a platform engineer and an SRE ?

A platform engineer and a site reliability engineer (SRE) are similar roles that both focus on building and maintaining the underlying infrastructure and platform that enables software applications and services to run effectively, but they have different emphasis and responsibilities.

A platform engineer is primarily responsible for designing, building, and maintaining the underlying infrastructure that supports the software applications and services. They focus on creating and managing the infrastructure, tools, and services that are needed to deploy, monitor, and scale software applications. They also work closely with development teams to ensure that the platform meets the needs of the organization.

On the other hand, an SRE (Site Reliability Engineer) is a role that emerged from the intersection of software development and operations, it’s a discipline that emphasizes the robustness, availability and scalability of software systems. SREs strive to improve the reliability, scalability, security, and performance of the software systems they operate, and they are responsible for the ongoing operation and improvement of systems, as well as the overall reliability and availability of services. They also have a strong focus on automation, monitoring and incident management.

In summary, a platform engineer is more focused on the underlying infrastructure, tools, and services needed to deploy, monitor, and scale software applications. While an SRE focuses on the reliability, scalability, and performance of the software systems they operate, and they are responsible for the ongoing operation and improvement of systems, as well as the overall reliability and availability of services.


End of Chat.

It’s very good. Interesting times ahead. I think it’s a better and more convenient form of gaining insight into a topic than doing Google Searches. One can use ChatGPT to gain knowledge first and then maybe dig deeper via google search results.

Does the tech industry need cadet programs similar to the army and the bureaucracy ? Is this a possible solution to the great resignation and the tech industry skilled worker shortage ?
In many countries at a young age cadets enter into the army with the view of a long term engagement. They are sponsored initially, trained a while and then they deliver later on. The important thing to note is that after all this they are committed in writing to serve multiple years.
Would it be fair to say that with the skilled workers shortage in the tech industry there is a requirement of cadet programs in the tech industry? Cadet programs where intake is made with a long term commitment. Where fresh people are taken in, skilled up and they serve a lengthy while ? This solves both the skills shortage and the employee churn in the current great resignation environment.
The relationship can be mutually beneficial for both sides just like it is in the army. A non-tech industry worker might be looking to enter into the tech industry while continuing to earn. The tech industry are also looking for people who can deliver. A six months or a year long on job training program with a view to start delivering on some tasks quickly and then develop further later. Again importantly the technology cadet program would have a proper written commitment to remain in the organisation for a couple of years at least. I think people looking to enter the tech industry will not mind such a commitment because of the remote work attraction in the tech industry. Its mutually beneficial.
This would convert every local McDonald’s worker and every Call Center Agent into a potential technology cadet available for intake. Anyone and everyone willing to enter the tech industry becomes an easy candidate.
I think its an option that should be available in the market.
What do you think ?

Analyst firms like Gartner have given indications that soon there will be digital assistants, digital employees, digital co-workers and/or advanced virtual assistants in the workplace. All these terms signify a similar trend. A technological bot like tool as a full co-worker. I will use the term Advanced Virtual Assistant.


In this write-up I intend to connect two phenomenon and give an outline of what an advanced virtual assistant for IP Networks might look like. The first phenomenon is Hey Google, Apple Siri and Amazon Alexa. These are personal virtual assistants which listen and respond to inputs from people. The second phenomenon is the open space available for a virtual assistant helping run IP networks and technology systems.


The basic example of a virtual assistant is that of Siri, Hey Google and Alexa. These are devices which listen and respond like a human. They require computation and processing to listen and calculate what is being said. Thereafter they require computation and processing to formulate a response and send it through. The I/Os are a mic to listen, a speaker to speak and communications over the Internet.


The way I see it is that every large enterprise and large network which is run by multiple network engineers could have a trained virtual assistant. Some of the basic repeated tasks of a Network Engineer are:

  • Run a ping or a traceroute on an IP
  • Login to a switch, router or firewall and run some commands
  • Look up an ISP circuits details from the ticketing system.
  • Look up routes in a router for a subnet
  • Look up empty IP address space in the IPAM. IP Address Management system.
  • etc. etc. etc.
    These are repeated actions regularly performed again and again by all Network Engineers throughout the world.

The question is: How could an Advanced Virtual Assistant help in these repeated actions?


From here on imagination starts. Imagine being able to speak to a bot on your work laptop. Lets call the virtual assistant Netty. Imagine being able to say Hey Netty, run a traceroute for 12.2.2.2 or Hey Netty, please you login to the route isbpk56. Imagine thereafter getting Netty to do this on the router: Hey Netty, run the command ‘show ip route 2.2.2.2’, or something like Hey Netty, run ‘show inventory’.


So basically, you have a Hey Google, Siri or Alexa type assistant for your network. It has been fed the networks IP addresses, limited rights login details, router names and other details as required. Now you are talking to it and it is doing some work. This is my understanding of ‘digital co-worker’ and ‘advanced virtual assistant’. An assisting device which you can talk to and it is able to execute the actions.

I feel that instead of large virtual assistants covering big domains there will probably be industry specific and audience specific virtual assistants. Even probably they will be company specific, trained and modified per organisation.


Let’s keep imagining. Imagine having a search option as well. For example: Hey Netty, search the network for any firewall rules on 10.62.45.3. This would be very handy when looking for some firewall rules. Or Hey Netty, search the network for the mac address. abcd.efds…


Similarly imagine getting a complaint for a circuit and saying Hey Netty, please bring up the circuit TX459’s details on the ticketing system. Or Hey Netty, search the network for the interface descriptions having the circuit TX459 in them.


I think this has a great potential to improve efficiency and productivity. If Network Engineers get a Hey Netty for their networks, they would probably be able to handle a lot of work very easily.


As expected, this will require work to get a conversational bot which is a good bot listener and a good bot speaker. There will also be search and response algorithms, databases queries and responses. etc. etc. That will be the work done to make Hey Netty possible. But it is an engineering exercise which would give much productivity increase especially for large networks run by large teams.


Similarly, the same is the case for Storage and Servers technology domains or even Telecom networks. The Storage engineers will have their own regular instruction sets, the servers and systems their own and telecom network their own. They could all get bot like virtual assistants trained to carry out actions.


Imagine saying this and getting the work done:
Hey Techy, could you show the details of the VM sydau78
Hey Telly, could you bring up the RF stats for the site melau86
Hey Netty, could you ping to the ip 3.3.3.3


I feel the industry leaders should step in to bring this into open standardisation efforts. So, for example Storage, Networking, Systems, Telecom industry leaders should each formulate teams to make standard Hey Netty like query-response lists. Thereafter bots could be trained for Hey Netty v1 and Hey Netty v2 as the query-response details come in from the standardisation body.
It is time for Hey Techy and Hey Netty to hit the technology space. It is time for the post-pandemic knowledge workers to also get advanced virtual assistants and digital co-workers. I feel this is a good time for standardisation efforts for these as well.


Thank you,
Syed Habib Kakakhel

A question very important is what will happen in the world when bandwidths become very high. Initially there was GSM call and DSL, then came 3g and fibre and now there are higher bandwidth fibre and 4g. There is 3g/4g and Wifi based fibre broadband in most places now.

To track the effects of bandwidths we can take a back to basics approach. It is that humans have eyes and ears. So it boils down to sight and speech. Therefore we can say that very high bandwidths like 100 Mbps in telecom and 1 Gbps on Wifi broadband will result in better things we can see, we can hear and talk to.

I feel 3D VR, AR and 4k, 8k video covers sight. In terms of speech and sound I feel voice operated devices and devices speaking interactively will become possible. Like Alexa, Siri and Hey Google.

Humans are limited in terms of seeing and hearing. Perhaps touching and neurological signals can be included but for now the eyes and ears are the primary interfaces that humans use to interact with technology.

Lets now take it further. With greater bandwidth and greater processing comes greater life-like responses from computers. So we there could be voice assistants assisting humans with all sorts of productive capabilities. Imagine you have a secretary and you can ask them to do stuff. Imagine now that there is a computer that is processor and bandwidth supported making it equivalent to a secretary and even more efficient. If one wants to schedule a taxi ask the device secretary, if one wants to schedule a flight ask the device secretary, if one wants to buy grocery ask the device secretary, if one wants to order something ask the device secretary. These are being called Virtual Assistants (Gartner predicts that, by 2025, 50% of knowledge workers will use a virtual assistant on a daily basis, up from 2% in 2019). So everyone will have a mini secretary in their smartphones and people who want to pay more will perhaps have sophisticated virtual assistants.

That’s ears, voice and speech.

Lets discuss eyes and sight. With greater bandwidth and greater processing comes the ability to show 3D life-like views to people (if they wear a headset and enter that domain). So perhaps a 3D camera will take in a 3D view of a meeting room and given sufficient bandwidth it will transfer that 3D view to another place. Someone wearing a headset can see a 3D view of any place and will really feel like he is there. Voice can be part of the transfer so people could wear a headset and be seeing and hearing a 3D view of a far of place. Instead of a 2D video call it’ll be a 3D view call. This makes way for new avenues in human interaction. So imagine an MIT classroom having a 3D camera and anyone in the world attending the class in 3D view via a headset. It will feel very much like a real class attendance and will definitely help learning. I think it will make a much better learning experience than the 2D video one because of its feel and presence effect. Its easy to get distracted into something other than what you are seeing on your 2D screen but a 3D view would be immersive learning. This is being called immersive learning and experiential training because it is immersive in experience and is not just sight but is an experience.

Enter The Matrix. You wear a headset and someone else wears a headset and they could be in a 3D virtual world together. Lets not make them fight here on the blog but two people being together in 3D view is a very real possibility given sufficient bandwidth and processing.

So now lets break this down. Consider a 3D live transmission and people attending in 3D view that live transmission. As in an MIT Class which people across the world attending in 3D view LIVE. Now separately consider non-live transmission. A virtual 3D world which 2 or more people are present in together which is a recorded experience. For example a recorded Safari trip which is not live but people jointly attend it. Now separately consider a computer generated 3D world in which there is a virtual assistant like a device secretary present as well. Someone wearing a headset is taken to a computer generated 3D virtual world where they are with a virtual assistant, are talking to them and learning from them etc. Now further consider 2 or more people being in a computer generated 3D world together and there being a virtual assistant present in that world as well which they can speak to. Interesting possibilities.

These are real possibilities given sufficient bandwidth and processing. I feel 5G and Wifi 6 causing bandwidth increase can result in these. Simultaneously edge computing and cloud computing causing increased processing can bring these things to life. These seem like real possibilities and things are progressing in this direction.