AI Infrastructure Project: the USA

The USA has announced an ambitions $500 billion investment in AI infrastructure. It will be a private sector investment aiming to take a lead over the rival nations in the business-critical technology.

It will be a joint venture of OpenAI, SoftBank and Oracle. The venture will be called Stargate.

Its project will be to build infrastructure of data centers and create more than 1 lac jobs in the US. The first project is under construction at Texas.

Masayoshi Son, Sam Altman and Larry Ellison launched the project from the White House in presence of President Trump.

The project will have 20 such infrastructure data centers spanning half a million square feet each.

The ultimate aim is to build artificial general intelligence (AGI). Microsoft has already collaborated with OpenAI for a $100 billion data center with an AI supercomputer also called Stargate.

AI does require leave capital expenditure in computing power (pushing demand for specialized data centers) to link thousands of chips together in clusters. They should also produce a lot of electricity, by setting up captive plants if necessary.

Stargate is a part of US-China tech battle. India should also look at foundational AI development. India’s reliance on US technology is evident but it will never adopt Chinese AI application programming interface.

Luo Fuli, Star behind the Star Performer: DeepSeek

DeepSeek is the pathbreaking Chinese model that rivals the performance of ChatGPT, Gemini and Claude AI. It tops the charts of App store — Apple. It has spread ripples through the stock market. It has been developed by a young and talented team.

One member of the team deserves special mention — Luo Fuli, a girl 29-year-old, hailed as AI Prodigy in China. She is known for her pioneering contributions to natural language processing (NLP). She has read at Beijing Normal University and has specialised in computer science. She later got a seat at Institute of Computational Linguistics (Peking University) where she published 8 papers. Alibaba and Xiaomi collaborated with her. She developed VEO, a multiligual pretraining model. She joined DeepSeek in 2022. She played a major role in developing DeepSeek-V2. It rivals ChatGPT.

Lei Juan, Xiaomi founder, offered her a compensation package of 10 million yuan per annum.

DeepSeek

In a matter of weeks in Jan 2025, DeepSeek has become the cynosure of the tech industry. Since ChatGPT was launched in November 2022, the tech industry realized that the Chinese model DeepSeek has managed to train a foundation model to rival ChatGPT and Llama at much lower costs by using fewer GPUs. It was a damp squib on the higher demand for advanced chips for the American models.

DeepSeek was started by hedge fund manager Liang Wenfeng in 2021. He purchased Nvidia chips for an AI side project while running his trading fund. He wanted to leverage AI to identify patterns that could affect stock prices. It was later that this project got converted into a standalone AI venture. In the training process, the model used rounding to make calculations easier. The cloud capacity was reconfigured to ensure the model splits tasks over multiple chips more efficiently.

DeepSeek has made most of its training results public and made the model open source.

The stock prices of tech stocks fell across the markets from New York to Tokyo. Maybe, it is an overreaction to the new model, which has the potential to lower the cost of developing an LLM. DeepSeek achieved inference at a much lower cost. This lowers the entry barrier for nations that would want to develop their own models.

Nandan Nilkeni is not in favour of India focusing on building LLMs. Arvind Srinivas, Perplexity AI does not agree. He expects India to build both LLMs and use the existing LLMs to build wrappers on top of them.

US president Trump has cautioned the US tech to take DeepSeek as a wake-up call.

One thing is certain. The Chinese so far emphasized only on making money. They ignored innovation. Innovation comes from a deep sense of curiosity and a desire to create. It is not always business or profit -driven. DeepSeek is taking China in the right direction.

Running a Restaurant

Restaurant business is a dicey proposition. Many new restaurants are set up, but many of them close down before the end of the first year. Some few survive and flourish, but very few get past the five-year mark. It is necessary to learn some lessons if your restaurant has ambitions beyond being fast food outlet of pizzas and burgers.

First of all, it is a daunting task to enter a market full of talented competitors. It is a cut-throat competition. Your restaurant must make the guests feel welcome when they walk in and order their food. It is necessary to maintain a good kitchen with various stations. Some restaurants allow guests to visit the kitchen to feel confident about the hygiene and cooking methods. There are restaurants which have a see-through transparent kitchen where the chefs can be seen cooking, at times showing their expertise by tossing the omelets up in the air and getting it down expertly in the fry pan. Kitchens must have good inventory management — the provisions must arrive in time. There should be quality control over the ingredients that go into the making of a dish. The accounting should not be in a mess — the costing of each dish on the menu and the surplus it generates must be clear. Manpower selection must be meticulous. It should be a well-composed team — captains, waiters, relationship managers, chefs, chief chef. There should be good division of labour. There should be control over things, and good control is exercised by proper delegation to the right people.

All the nitty gritty of running a restaurant is okay, but as you are a part of the hospitality industry, you should be recognised by your peers. They should welcome you, though they are your rivals. The managers must be affable, rather than hostile. It is smart diplomancy that is required to run a restaurant. In the neighborhood, a restaurant must earn tonnes of goodwill.

Abroad, a restaurant takes pride in earning Michellin stars and would love to be a good ranked restaurant. It is necessary to maintain the stars once you have earned it — a restaurant with three stars can fall to two. There is stress of maintaining the status quo.

It is nice to be a well-funded restaurant of a group. At the same time, you should also be aware that such a group can take hasty decisions of revamping the kitchen or replacing it , and changing the layout at the drop of a hat. They can change the cuisine to shore up their bottom line.

You can run a small low-key restaurant lifelong. Still, these days it is necessary to be in the good books of regular or social media. It ensures a steady flow of diners to your establishment.

You make people happy by being nice. That itself is hard work. Happy people create a feedback loop.

Agentic and Physical AI

Human beings, through the course of civilization, strived to augment their physical strength by using tools and machines. First, they harnessed the power of animals, and later used the power of steam, electricity, and nuclear reactions. Human muscles were replaced by machines. We are witnessing in the current year (2025) the emergence of the AI agents. These augment our cognitive abilities, rather than the physical power. Thus, it is an inflection point.

These days there is a convergence of agentic AI, intelligent software agents capable of working autonomously and physical AI, consisting of robots and machines.

The agentic AI are going to transform the way we work, live and interact. They will transform the workplace and office. They will democratize access. Anyone, regardless of digital proficiency, will be able to interact with AI agents as these will be multi-modal. There will be a network of AI agents who will act as intelligent assistants. They will free us for higher level functions.

AI will be extended to the physical world, e.g. robots. It will unlock new frontiers of automation and efficiency. There are going to self-driving vehicles, and autonomous drones. There will be AI-powered infrastructure for smart cities. The healthcare delivery system will change.

AI agents will unlock hidden opportunities which we have not so far exploited.

Net Neutrality

Net neutrality has remained a topic of debate. Recently, a federal appeals court has struck down the net neutrality rules of the Federal Communications Commission (FCC).

In 2015, FCC has issued Open Internet Order — it guaranteed net neutrality. As a principle it meant that internet service providers will treat all communications equally. This results into consistent service and pricing for both users and content providers. In 2017, FCC rescinded the net neutrality requirement. It was reinstated in 2024 by President Biden. In January 2025, it has been struck down again.

The users of internet have not experienced any changes at all.

To begin with, it was argued net naturality was essential for freedom speech and freedom of the inherent. If there is no net neutrality, the small corporations would decide the content and would charge more for the content they found objectionable or unprofitable. Some even contended that one would have to pay for internet one word at a time. The reality is vastly different. There is expanded bandwidth. There is plenty of access. Neutrality or no neutrality, it has remained fully accessible.

The supply is elastic. The sky is falling mentality is not good. It was argued net neutrality is required to bridge the digital divide. It was predicted that in the absence of net neutrality, the streaming business would crash. Introduction of net neutrality was hailed as historic, and end of net neutrality as a move towards internet’s death. Such commentators do not appreciate the economics. Internet in the 1990s flourished under non-conformist anticorporate spirit.

Today, there are complaints about the harmful effects of internet on minors and excess of pornography. It means there is excess of access to the internet. It is just the opposite of what the supporters of net neutrality warned about. Markets have expanded output.

One lesson from this debate is about free speech. Those who control and transmit information across the internet are essentially calling for censor to limit the speech of some users.

Net neutrality — good bye! And who will miss it? None.

Genome India Project

Under the Genome India Project, the Department of Biotechnology announced a new platform and framework for sharing human genome dataset duly sequenced. It consists of 10,000 genome sequences of healthy individuals from 99 ethnic populations in India. It will help create a baseline map of India’s genetic diversity. In the second phase, the researchers will sequence genomes of people with specific diseases.

Genomes and Their sequencing

We inherit from the parents an instruction manual that decides how our body develops and functions. Right from our height to the colour of our eyes and hair is decided by the instruction manual. It also decides the diseases we inherit or are pre-disposed to. The name for this manual is genome which is made up of twisted paired strands of deoxyribonucleic acid or DNA.

Each strand of DNA consists of four chemical units, or bases, denoted by the letters A,C,G and T. The order of these bases determines the meaning of information encoded in the DNA. There are various combinations of these bases (say some three billion pairs in the complete human genome). These determine the complete genetic makeup of each individual.

What is sequencing? Sequencing refers to the exact order of theses bases in an individual strand of DNA. This is done by extracting information from the blood. Every single cell virtually contains a complete copy of the individual’s genome. However, it is difficult to handle the entire genome.

A genome is cut into smaller pieces and these pieces are tagged. They are then decoded. As it is, they smaller chunks of genetic material. The information is reconfigured using tags to create the whole genome.

India Project

India has diversity — some 4600 distinct populations. It leads to genetic diversity. The Genome India Project seeks to capture this diversity. Under this project, 10,000 genomes will be sequenced.

Ultimately, one million genomes will be sequenced.

Benefits

It will be easy to identify genetic basis of diseases and the risks for such diseases. The therapies and doses can be targeted. The new diagnostic tests could be developed.

Genetic manipulation — modifying, deleting or adding certain genes — can lead to new therapies and can eliminate such diseases.

There are 135 million genetic variations in 10,000 genome sequences. Some 7 million variation are not found in other global databases. We will thus identify specific Indian genetic variations. Scientists will identify their frequency, to assess how it leads to a common disease. MYBPC3 mutatron leads to cardiac arrest at young age. It is found in 4.5 per cent of Indian population but is rare globally.

We will also identify resistance-inducing variations.

Certain medicines and anesthetics are ineffective in certain populations.

Second Phase the Project

Here genome sequencing will be done of people with specific diseases. A comparison can be made between a healthy and diseased genome. It will help identify the culprit genes that cause the disease or predispose to that disease.

Researchers will also identify genetic changes induced when one suffers from a particular disease.

Sequencing will be done for different types of cancer, chronic conditions such as diabetes, various neurological or neuro-degenerative diseases and certain rare diseases found in Indian population.

The data will be protected and care will be exercised while sharing the data. The data could be shared with institutes who parther with us in the project.

The data will be double-blinded to maintain anonymity. Samples from partner institutes will be encoded before uploading and will be encoded again when shared with researchers.

AI Application

AI is acting as a transformative force across industries. It challenges the traditional business models and unlocks new possibilities.

Most of the organisations use AI in customer service. AI chatbots with robust capabilities improve customer interactions. Another area of AI application is employee services. AI helps employees to access information on benefits, policies and other resources.

The next big application of AI will be in its ability to process and analyze massive data sets. It will play a key role in addressing climate change and sustainability challenges. AI can optimize the movement of goods, people and traffic and could make our cities smarter. Most promising uses of AI will be in the area of health care — developing new treatments and transforming patient care.

The next phase of AI is already unfolding. It consists of agentic frameworks. Autonomous AI agents will perform specific tasks. Organizations will leverage AI agents to transform operations.

The sectors where AI will play a key role are financial services, healthcare, media and telecom. AI may affect certain sectors negatively to begin with — accounting and legal services. These industries will have to adapt and reinvent themselves with AI-powered tools.

Organizations will have to be ready for AI-driven transformation. There should be reskilling of the manpower. Reskilling should be a collaborative effort of the government and organisations.

Some may misuse the AI technology. Therefore, there should be emphasis on ethical business practices. There could be deepfakes. We will have to develop more sophisticated methods to verify and validate the information we consume.

Operator: OpenAI’s AI Agent

OpenAI begins the new year 2025 by introducing in January its general-purpose AI agent that could take control of the web browser and independently perform certain actions. To the US users, it will be available on subscription plan first and will be introduced to the other countries soon. Later it will be integrated to all its ChatGPT clients.

The tasks Operator could perform are: booking travel accommodations, making restaurant reservations and shopping online

The agent will use a dedicated web browser to complete the tasks. Users can still take control of the screen as the Operator is working.

Operator is powered by CUA or Computer-Using Agent model. It blends the vision capabilities of the GPT-4o with the reasoning abilities of OpenAI’s advanced models. CUA is trained to interact with the front-end of the web-cites (it does not have to use APIs to tap into different services). It can use buttons, navigate menus and fill out forms on a web page just as humans do.

OpenAI collaborates with eBay, Instacart, StabHub, Uber, Priceline, DoorDash so as to ensure that Operator respects the terms and conditions of those businesses.

The CUA model is trained to ask for user confirmation before finalizing tasks. A user can double-check the model’s work before it becomes final.

OpenAI warns the CUA is not perfect. It is not reliable across all the scenarios. OpenAI recommends an abundance of caution and supervise certain tasks. (such as banking transactions). Users may have to intervene for card transactions. Operator does not.

There are some limitations of the Operator. There are rate limits – both daily and task-wise. It denies certain tasks for security reasons. This can change in future. It also gets stuck if the task is complex. It will ask the user to take over.

OpenAI is slower than others to develop agents. It considers the safety risks around the technology. It considers Operator safe enough. It is a bold attempt.

New AI Paradigm

Yann LeCun, Facebook-Chief Scientist AI, predicts a new paradigm of AI architectures likely to emerge in the next three to five years, which will have the capabilities far beyond the present systems.

LeCun also predicts that the coming years could see emergence of robotics combined with AI to unlock a new class of applications.

LeCun spoke at Davos in January 2025 in a session called Debating Technology. He feels the present generative AI and LLMs, though useful, fall short of the expectations. The new architecture will overcome the limitations of the present system which lack the understanding of the physical world, lack of persistent memory, lack of reasoning and lack of complex planning.

Another AI revolution is in the offing. Maybe, it will acquire new nomenclature as the present nomenclature of generative AI will not be appropriate for it. At the heart of this technology are the world models which will make the machines understand the dynamics of the real world. The qualities the new models would possess are persistent memory, common sense, intuition, and reasoning abilities. The current systems have not gone beyond pattern-recognition. Though LLMs process natural language, they are not good at thinking. If the research succeeds, they will have a different paradigm.

In robotics, the focus will be on general purpose, adaptive and versatile robots with intelligence equaling human intelligence. The coming decade could be a decade of robotics.