Who will win the race to AGI?

AI landscape at present

When it comes to leadership in AI, Google has maintained its leadership owing to its early investments in machine learning and expansive infrastructure. However, in the last few years, Sam Altman-led OpenAI came up as a worthy contender to the Alphabet Inc. company. OpenAI was able to catapult itself in the global race, courtesy of its top-of-the-line GPT models including GPT-4 Turbo. The AI startup’s partnership with Microsoft has also given it an edge, since Microsoft’s Azure platform is a key distribution channel for AI services. While OpenAI’s profitability is a contentious matter, the company has reportedly raked in billions of dollars in revenue from API access and its enterprise solutions. 

Meanwhile, Google has utilised its massive ecosystem of search advertising and cloud services to offer AI features across its products and services. Meta has also made some significant strides with its LlaMA models, however, they are not reportedly generating any direct revenue. Instead, the social media giant reportedly profits through its AI-driven recommendation systems across its platforms. 

Financial dynamics of AI

AI developments are drawing money from all quarters. A look at recent earnings reports of big tech would reveal their aggressive investments in AI infrastructure. Amazon, with its massive $100 billion in capex in 2025 for AI workloads on AWS, leads the pack. Microsoft follows close with an $80 billion commitment for data centres, while Alphabet is allocating $75  billion to expand its AI initiatives. Meanwhile, Meta is planning an investment of up to $65 billion for computing infrastructure. However, in all of this, Apple seems to be taking a cautious route. All these spending sprees are happening at a time when there is a growing skepticism owing to the rise of Chinese AI startup DeepSeek whose AI models challenge conventional resource-heavy AI development. 

On the other hand, when it comes to profits, translating innovation into monetary gains seems to be a persistent challenge. Microsoft generates a significant revenue from its AI products, but has high capital expenditures on GPUs and infrastructure. Similarly, Meta reportedly earns from recommendation algorithms, but it also loses money on its LlaMA models. On the other hand, Nvidia is the clear hardware winner, as it is raking in billions of dollars from AI-related GPU sales. 

The Deepseek logo is seen in this illustration taken on January 29, 2025. (REUTERS/Dado Ruvic/Illustration/File Photo)

In contrast, OpenAI and Anthropic are reportedly not yet profitable and continue to rely on funding from external sources. It needs to be noted that the cost of training advanced AI models continue to be exorbitantly high, constantly requiring heavy investments in research and compute resources. There is no official figure, but the model development cost of OpenAI is estimated to be in the hundreds of millions of dollars. 

Another major challenge faced by AI companies is the potential of commoditisation of AI models. With open-source alternatives becoming cheaper, the value of proprietary models are likely to decline. Companies like DeepSeek pose a threat to OpenAI, whose primary value proposition is its proprietary models like ChatGPT. It seems the future of AI revenue will move above simple chat use cases. There will be more focus on code generation, agents, robotics, and task automation. In case AI becomes a commodity, companies that will know how to effectively advertise within AI outputs, may hold the edge over others. Google AdSense is an example of what all can happen with this kind of monetisation. 

The AGI investment gamble

Most of the competition in AI is accelerated by the idea that AGI may be only a few years away. There are research labs and companies that believe AGI could be built within two to three years, while many continue to maintain that it would have a five to ten-year timeline. In a post on his blog, OpenAI CEO Sam Altman said that OpenAI is confident and knows how to build AGI. He went on to predict that in 2025, the world may see the first AI agents join the workforce.

This expectation of achieving AGI is also shaping the investment strategies, as companies are racing to amass the necessary resources for the breakthroughs. AGI is not just about training models; it requires extensive research, especially in areas such as compute efficiency and reasoning. From what we are observing, companies like OpenAI and Anthropic are laser-focused in this direction. They also strongly believe that achieving AGI will unlock massive financial gains. 

While companies like OpenAI and Anthropic rally behind their state-of-the-art models, bigger companies like Google, Meta, and Amazon have an infrastructural edge over them.These companies can use AI to enhance existing products and services. And even if they do not build foundational models, they can still benefit from AI’s overall advancement.  This also diversifies their revenue streams and they may not have to depend on a single AI application. 

The wild card entry

January 2025 showed the world that China’s aspiration with AI cannot be undermined. Over the last few years, China has positioned itself as a major player in the development of AGI. DeepSeek is one of the many AI startups in China that is harbouring a grander vision of an AI-powered future. The socio-political environment, significant investments, and government push are fostering innovation here. Moreover, many experts feel that the nation’s centralised governance allows swift decision-making and a hassle-free integration of AGI across sectors – from public surveillance to economic planning. China has no constraints of democratic processes, and its loose data privacy regulations offer tech companies a vast dataset needed for training AGI. Despite the ongoing semiconductor restrictions, the Chinese government is spending heavily in AI research and development, solidifying its tech leadership.

On the other hand, the United States of America, under President Donald Trump, has announced the Stargate Project, a $500 billion initiative to build AI infrastructure in the country. The US tech companies are aggressive in their pursuit to bring the most advanced AI, and China holds many surprises, making the race intense. 

US President Donald Trump delivers remarks on AI infrastructure, next to Oracle co-founder Larry Ellison, SoftBank CEO Masayoshi Son and OpenAI CEO Sam Altman at the Roosevelt room at White House in Washington, U.S., January 21, 2025. REUTERS/Carlos Barria

Who will take the lead?

The race to achieve AGI is growing more intense, and as of now, it seems unlikely to be a winner-take-all scenario. OpenAI and Anthropic, considering their operations, may run into long-term sustainability issues unless they don’t diversify their revenue models. On the other hand, Google, Meta, and Amazon have the advantage of integrating AI into their existing ecosystems. Ultimately, it seems that a company that can manage to balance innovation with sustainable monetisation might emerge as the leader. The race continues now with each player betting on their unique path to the future of AI.