{"id":227547,"date":"2025-01-30T08:04:27","date_gmt":"2025-01-30T08:04:27","guid":{"rendered":"https:\/\/news.talkwithrattan.com\/index.php\/2025\/01\/30\/india-vs-china-why-country-of-techies-cant-create-homegrown-deepseek-or-google-entrepreneurs-reveal-whats-holding-india-back\/"},"modified":"2025-01-30T08:04:27","modified_gmt":"2025-01-30T08:04:27","slug":"india-vs-china-why-country-of-techies-cant-create-homegrown-deepseek-or-google-entrepreneurs-reveal-whats-holding-india-back","status":"publish","type":"post","link":"https:\/\/news.talkwithrattan.com\/index.php\/2025\/01\/30\/india-vs-china-why-country-of-techies-cant-create-homegrown-deepseek-or-google-entrepreneurs-reveal-whats-holding-india-back\/","title":{"rendered":"India vs China: Why country of techies can’t create homegrown DeepSeek or Google? Entrepreneurs reveal what\u2019s holding India back"},"content":{"rendered":"
\"India<\/div>\n
In a recent social media discussion, Sridhar Vembu<\/a>, Chief Scientist of Zoho Corporation<\/a>, responded to a frequently asked question from Indian engineers: \u201cWhy can\u2019t India achieve what China<\/a> did?\u201d Vembu acknowledged that while India\u2019s strengths lie in its democratic framework and diverse society, it must chart its own path rather than merely emulating China. He emphasised grassroots innovation and India\u2019s demographic advantage, asserting, \u201cIndia’s strength lies in its ability to innovate and adapt to rapidly changing global dynamics.\u201dThis conversation has gained urgency in light of China’s latest AI breakthrough: DeepSeek<\/a>. The Chinese AI model has stunned global tech circles, outperforming OpenAI<\/a>\u2019s models, topping the iOS App Store, and displacing Meta<\/a> in the open-source AI space\u2014all at a fraction of the cost. DeepSeek\u2019s success has reignited concerns among Indian tech leaders: Why isn\u2019t India leading in AI?<\/p>\n

<\/p>\n

What Makes DeepSeek So Significant?<\/h2>\n

DeepSeek R1, a large language model (LLM) comparable to OpenAI\u2019s ChatGPT, reportedly cost only $6 million to develop, whereas OpenAI spent $540 million on ChatGPT. Moreover, DeepSeek uses far fewer GPUs, significantly reducing operational costs. This frugal innovation, reminiscent of India\u2019s own Chandrayaan-3 space mission, has shaken global markets.<\/p>\n

Following the release of DeepSeek R1, American tech giants\u2014including Nvidia<\/a>, Meta, and Alphabet\u2014lost billions in valuation as investors scrambled to reassess AI\u2019s economic landscape. The disruption has underscored China\u2019s growing dominance in AI and renewed debates over why India hasn\u2019t produced similar breakthroughs.<\/p>\n

<\/p>\n

Why Has India Lagged Behind in AI?<\/h2>\n

Zerodha founder and CEO Nithin Kamath weighed in on the discussion, arguing that India’s technological struggles are deeply rooted in systemic issues. \u201cIn the 1960s-70s, India and China had roughly the same per capita GDP. They started their reforms in the 1980s, and by 1990, they had overtaken our per capita GDP. Say what you will about the differences in our worldviews and economic models, but their scientific and technological progress is undeniable across disciplines\u2014DeepSeek is just the latest example.\u201d<\/p>\n

\n
\n

\u2014 Nithin0dha (@Nithin0dha) <\/p><\/blockquote>\n<\/div>\n<\/div>\n

Kamath pointed out that India suffers from a culture of short-termism, where challenges are often addressed with patchwork solutions rather than long-term strategies. \u201cYou can\u2019t just buy GPUs and expect Indians to create groundbreaking AI applications. Without the right talent and an enabling ecosystem that facilitates innovation, all the GPUs in the world will be pointless,\u201d he explained.Additionally, India does not provide an adequate environment for researchers. \u201cWhile India does produce great researchers, we don\u2019t seem to offer a conducive environment for them, which is why the majority of them go to the US,\u201d Kamath stated. He stressed that research takes time and sustained investment, noting that China spent over two decades systematically building its AI and scientific capabilities.<\/p>\n

The Hard Part of AI Innovation<\/h2>\n

According to an anonymous AI researcher at Amazon<\/a>, known online as GDP, the fundamental science behind LLMs is not particularly difficult. They explained that \u201call these models are Transformer Decoder-only models, an architecture that has existed since 2017.\u201d Many of the key improvements in AI\u2014such as Flash Attention, ROPE, Mixture of Experts<\/a> (MoE), and GRPO\u2014are open-source and relatively straightforward to implement.<\/p>\n

However, the real challenge lies in large-scale AI training. \u201cIt is the parallel and distributed computing to run AI training jobs across thousands of GPUs that is hard,\u201d GDP noted. DeepSeek excelled in this domain by using innovative techniques to optimise computational efficiency, allowing them to achieve remarkable results with minimal resources.<\/p>\n

India\u2019s Missing Piece: A Protected Market and Patient Capital<\/h2>\n

One of the biggest reasons why India hasn\u2019t built an AI model like DeepSeek is the lack of a protected market. In China, domestic tech firms are shielded from foreign competition, allowing them to experiment, fail, and ultimately succeed. \u201cYou need to be able to walk before you can run. There is no protected market to practice your craft in early days. You will get replaced by American service providers as they are cheaper and better every single time,\u201d GDP pointed out.<\/p>\n

In contrast, China treats AI development as a national security priority. Even when Chinese AI startups underperform, they continue to receive funding from government and corporate backers. This ensures that top-tier talent remains within the country, leading to a Darwinian competition where the best models eventually emerge victorious.<\/p>\n

A Call for India\u2019s Own \u2018DARPA\u2019<\/h2>\n

To bridge the AI gap, industry leaders are calling for India to establish a government-backed fund dedicated to AI research. GDP proposed the creation of an Indian DARPA\u2014a national programme with a $3 billion budget over three years, designed to fund AI research without immediate revenue expectations. \u201cWhat we need is a mentality that treats this skillset as existential. We need a national fund that will support AI teams and ensure they meet increasingly difficult benchmarks every six months. No revenue needed to survive for the first three years,\u201d they argued.<\/p>\n

Vembu echoed this sentiment, urging Indian engineers to embrace the challenge rather than blame systemic obstacles. \u201cChinese entrepreneurs never had it easy and still don\u2019t have it easy. We can do this in India. We WILL do this in India,\u201d he asserted.<\/p>\n

Can India Catch Up?<\/h2>\n

The emergence of DeepSeek has proven that groundbreaking AI models can be built with far less money and computing power than previously thought. It has also shown that small teams\u2014when supported by long-term investment and a nurturing ecosystem\u2014can compete with global tech giants.<\/p>\n

For India, the lesson is clear: Without sustained investment in research, a supportive environment for talent, and a long-term vision, the country risks falling further behind in the AI race. As Kamath warned, \u201cWith AI set to reshape the world, being mediocre won\u2019t be an option.\u201d
<\/p>\n<\/div>\n