Chapter 118: Open AI is not so Open, but Chin[A](I) might be?
18 Feb, 2025
(This cover image was suggested by DeepSeek and created with OpenAI's DallE 3.)
This week, Typeright takes a dig into the new AI contender and what it means for us.
The release of the more advanced Chat GPT 3.5 two years ago was quite a game-changer, launching what might be called a new 'arms race' between leading tech giants like Microsoft, Google and Meta to build the best AI LLM out there, pumping in several scores of billions of dollars and a not so tiny impact on the environment.
OpenAI, SoftBank, Oracle, and MGX have unveiled the Stargate Project, a collaborative initiative intended to allocate $500 billion into US AI infrastructure by 2029. The initiative will establish 10 data centers in Abilene, Texas, with potential extensions in additional states and countries. The program is anticipated to generate more than 100,000 employment and is endorsed by prominent technological partners such as Arm, Microsoft, Nvidia, Oracle, and OpenAI. The initiative seeks to preserve the United States' preeminence in artificial intelligence technologies.
2 GW is a lot of power. The report also reads that "Entergy recently proposed to develop a 1.5GW natural gas plant in Louisiana, near Holly Ridge, for a new data center customer. Reports linked the plans to the new Meta project." Zuckerberg's announcement post also mentions an investment of "$60-65B in capex this year while also growing our AI teams significantly, and having more capital for future expansions.
From a Forbes' article explaining the increasing spending on AI -
1) AI is expected to have a multi-trillion dollar economic impact globally, with a recent estimate from IDC placing AI’s cumulative potential impact through 2030 at $20 trillion. The mobile economy, which sprouted a handful of the trillion-dollar tech behemoths of today, added approximately $5.7 trillion to the economy in 2023. Big Tech’s leaders are well aware of how critical it is to capture and capitalize on an opportunity of this magnitude, and will not miss it.
2) Developing larger models and doubling model sizes requires massive computing power that only Big Tech can afford to develop, meaning a majority of genAI progress is likely to be made primarily in the hyperscalers’ clouds.
3) Big Tech is already realizing AI-related gains, with three of the four saying AI revenue is at least in the mutli-billion dollar range. With millions to billions of users for products to either enhance with AI integrations or target with AI features in subscriptions, the long-term revenue opportunity could dwarf some of their leading revenue streams of today.
The demand for power is so much that apparently, the companies are in talks to reopen previously shut-down nuclear power plants.
The Three Mile Island Reactor was in the process of being decommissioned after witnessing one of the most serious nuclear meltdowns back in 1979.
It is in between all this, a chinese start-up released the latest model of their own AI LLM, powered by fewer and lesser advanced chips, running on a lot less power and a fraction of the funds - and manages to rival the competition, even one-upping them at several tasks.
On asking the founder Liang Wefang why he built an AI model, he claims it's curiosity. "“From a broader perspective, we want to validate certain hypotheses. For example, we hypothesise that the essence of human intelligence might be language, and human thought could essentially be a linguistic process. What you think of as ‘thinking’ might actually be your brain weaving language. This suggests that human-like AGI could potentially emerge from large language models."
More importantly, the model is fundamentally open source, which OpenAI is not. Why did OpenAI choose to release an API instead of open-sourcing the models? From their blog:
There are three main reasons we did this. First, commercializing the technology helps us pay for our ongoing AI research, safety, and policy efforts. Second, many of the models underlying the API are very large, taking a lot of expertise to develop and deploy and making them very expensive to run. This makes it hard for anyone except larger companies to benefit from the underlying technology. We're hopeful that the API will make powerful AI systems more accessible to smaller businesses and organizations. Third, the API model allows us to more easily respond to misuse of the technology. Since it is hard to predict the downstream use cases of our models, it feels inherently safer to release them via an API and broaden access over time, rather than release an open source model where access cannot be adjusted if it turns out to have harmful applications.
The Open Source movement has existed for a few decades now and has effectively countered the myth of profit-motive being the best for innovation. Founded on the principles of Collaboration, openness and inclusivity, we have seen how open source software matches or fares better than the closed counterparts in terms of the security the openness provides.
But why is most of the world panicking? Well, we decided to go and ask the LLM itself what it thinks. Here was deepseek's answer:
American tech companies are increasingly concerned about the rise of DeepSeek, a Chinese AI firm focused on advancing general artificial intelligence (AGI) and deploying AI solutions across industries like healthcare, finance, and autonomous systems. This unease stems from multiple interconnected factors that threaten U.S. technological dominance.
First, DeepSeek’s progress in AGI—a field seen as pivotal to future economic and military power—signals a narrowing gap between Chinese and American innovation. China’s aggressive national strategy, including a $150 billion AI investment plan and state-backed collaborations between firms like DeepSeek and academic institutions, creates a coordinated ecosystem that accelerates breakthroughs. Unlike U.S. companies, which navigate stricter regulations and public skepticism around data privacy, DeepSeek benefits from China’s flexible regulatory environment, enabling rapid experimentation in sensitive areas like facial recognition and smart city infrastructure. This asymmetry allows Chinese firms to iterate faster and scale AI applications in real-world settings, such as mass surveillance networks, which American tech giants often avoid due to ethical and legal concerns.
Compounding this challenge is China’s vast population, which generates unparalleled volumes of data for training AI models. DeepSeek’s access to diverse datasets—from healthcare records to urban mobility patterns—grants it an edge in refining algorithms for global markets, while U.S. firms face fragmented data governance laws and limited access to such expansive, centralized pools. Geopolitical tensions further complicate the landscape, as U.S.-China trade restrictions on semiconductors and investment flows risk creating parallel tech ecosystems. American companies fear being locked out of China’s massive consumer market while DeepSeek expands into emerging economies in Asia, Africa, and Latin America with cost-effective, locally tailored AI solutions.
Talent competition adds another layer of anxiety: DeepSeek’s ability to attract top global researchers with lucrative incentives threatens to drain Silicon Valley’s innovation pipeline. Historical suspicions of Chinese intellectual property theft, though unproven in DeepSeek’s case, amplify fears that state-supported firms could leverage acquired knowledge to bypass costly R&D phases. Cybersecurity concerns also loom large, as DeepSeek’s perceived ties to China’s government raise questions about data sovereignty and espionage risks, making U.S. firms hesitant to adopt Chinese AI tools despite potential cost advantages. Finally, DeepSeek’s financial backing from state-linked investors allows it to absorb losses and undercut competitors through aggressive pricing—a tactic previously seen in China’s solar and electric vehicle industries—which could pressure American companies to retreat from key markets.
Together, these factors paint a daunting picture: DeepSeek embodies China’s strategic ambition to lead the AI revolution, leveraging state support, regulatory leniency, and vast data resources to challenge U.S. supremacy. For American tech giants, the stakes extend beyond market share; they involve shaping the global norms, security frameworks, and economic hierarchies that will define the AI-driven future. The response—whether through increased lobbying for government subsidies, cross-border alliances, or accelerated in-house innovation—will determine whether the U.S. can maintain its edge in a field where technological leadership is increasingly synonymous with geopolitical power.
Asking ChatGPT its comments on the same, here is one part of the answer:
At the same time, there are real concerns about the broader implications. DeepSeek’s open-source approach—while a boon for transparency and community-led innovation—raises questions about intellectual property protection, data privacy, and national security. The fact that its technology is developed and deployed under a regime with strict censorship and state influence has led to fears that sensitive data or strategic capabilities might be inadvertently exposed or manipulated. This is especially significant given that its rapid rise in popularity caused dramatic market reactions, including significant stock losses for companies like Nvidia
Now that we heard them, we wanted to unpack some of this.
First, the claim of intellectual property. OpenAI CEO Sam Altman has previously admitted that it would be impossible to "train today's leading AI models without using copyrighted materials." The same company now claims DeepSeek has used their work.
Let us not forget Aaron Swaartz, the young programmer, hacker and activist who was prosecuted for making available JSTOR articles, and ultimately led to suicide.
His life was cut short in 2013, after federal prosecutors charged him under the Computer Fraud and Abuse Act (CFAA) for systematically downloading academic journal articles from the online database JSTOR. Facing the prospect of a long and unjust sentence, Aaron died by suicide at the age of 26. (from EFF's page)
AI companies with stocks owned by the world's richest people, however are immune to all this.
More interesting is the geopolitical dimensions. The previous Trump Presidency went all out in initiating several trade restrictions with China, including their access to some of the US's higher end chips and technologies. The aim was to choke their innovation, but the results were surprisingly the opposite. China does not have the cleanest record of human rights and freedom of expression, as can be seen with any quesitons about the chinese government to DeepSeek. This is flagged by most news outlets - but similarly, the US based AIs have their own ways of navigating questions about Palestine, and Indian ones about difficult questions regarding their origin country. Where do we overlook these and aim at the larger realities of what advanced open innovation can do?
In Other News
First, some articles by DEF Founder and the Research Team:
DEF's Founder Osama Manzar quoted in an article- "The first concern is the contextual relevance of the AI’s responses. ChatGPT’s backend systems may not adapt well to the diverse linguistic and cultural needs of local communities, there are concerns about data privacy too.”
In today’s world, where everything is moving online, those without digital access are left behind—not just in education or opportunities, but in the most basic need of all: food. DEF's Arpita and Osama writes for TNM-
Also by the same authors - The progress of digitalisation in India has been steadfast with over 650 million smartphones users and over 950 million internet subscribers. Yet, the same country accounts for the largest number of unconnected citizens, exposing a deep social divide. This paradox raises questions about accessibility and equity in India’s digital journey.
Maitri Singh of DEF writes, "The path forward requires deliberate, well-planned interventions. Comprehensive assessments of socio-economic realities, inclusive community engagement, and tailored transition plans are essential. Policymakers must prioritise raising awareness about the impending changes and involving local communities in shaping solutions."
Join us for the launch of a research study by DEF India and Swaniti, exploring coal dependency in Ramgarh & Bokaro & pathways to sustainable development.
TypeRight - The Digital Nukkad, is a weekly conversational bulletin curated through the news and discussions on social media as well as what's happening on the ground. Through the eyes and ears of Digital Empowerment Foundation across rural India and global south, TypeRight aspires to focus on bringing the contextual relevance of digital technologies and developments on the society - both connected and unconnected.
Write a comment ...