Emerge 2024 Story of the Year: The Race to Rule AI

OpenAI's rise to tech stardom reads like a Silicon Valley soap opera. After Sam Altman's Amazing Return in 2010, By 2024, the $157 billion company that got off to a rocky start has transformed from a cautious nonprofit to an AI powerhouse. With a $13 billion investment from Microsoft and a deal to power Apple's iPhones, the company is on track to generate $11.6 billion in revenue.

Is it game over for all the other brats trying to elbow their way into the AI ​​market? Hardly.

Funding into AI continues to reach tsunami proportions in 2024, with billions of dollars funding dozens of worthy contenders around the world, from China's Moonshot AI to Paris-based Mistral.

In the last week of November alone, Anthropologie and Elon Musk's xAI pocketed $4 and $5 billion, respectively. Investors who once saw OpenAI as the next big thing in technology are hedging their bets on squeamish, hungry competitors and anything that ends up in AI could be the next frontier – especially after OpenAI's delays, including Sora, BallyHood's audio cloning tool (and, presumably, the next GPT model ).

It seems like it's been decades since Chatgpt burst onto the scene with its revolutionary chatbot, which users can interact with as easily as talking to a friend. (The chatbot launched in November 2022 and added speech in September 2024.) Overnight, Google search made something old-timers use look archaic. Nowadays, GenZ is probably the first thing that comes to mind when someone says “AI”.

The company that sparked the AI ​​revolution with ChatGPT in 2011. After the development of GPT-3.5 in 2024, pushing the limits of AI capabilities, in May the multimodal GPT-4o scored an unprecedented 88.7% on the MMLU benchmark. In September, the new o1 model – which is supposed to handle complex reasoning – topped the rankings again, scoring 83.3% accuracy on the International Mathematical Olympiad questions – a huge improvement over GPT-4o's 13.4%.

With more than 200 million weekly active users, OpenAI's impact was so profound that Google, plunged into the Innovator's Dilemma, suddenly began to fear for its $160 billion search business.

But money and fame don't always guarantee a happy ending, especially in tech. Anyone remember Blackberry?

Google is all about raw power – but people love its chats

Google certainly won't be going BlackBerry's way anytime soon. Besides, it's not the type that gets caught with its pants down. OpenAI launched in December 2015, and two months later, the search giant unveiled its new Gemini Ultra Foundation model with the capacity to process 2 million context tokens—making OpenAI's GPT-4 lightweight. AI is so important to Google that CEO Sundar Pichai has announced that the company is shifting from mobile to an “AI-first” strategy.

And not a second too soon: Switching to Gemini Google One—the tier that gives Gemini Ultra access—supplied more than 100 million subscriptions 24 hours after its release.

Google isn't just flexing its technical muscle; ChatGPT is still widely used but Google may be on to something. Its base model also leverages RAG, short for “Recovery Augmented Generation” platform Notebook LM, originally conceived to help people manage large volumes of data and large volumes of files. Product won't move the needle until an update changes the way people use RAG models. Welcome to Google Podcast Generators.

That feature alone was enough to boost the model's popularity, and it was enough to drive user engagement through a pretty active Reddit community, interesting social experiments, and some commercial applications.

“I think we learned a lot last year. What's resonating with people, what's really useful, how are they using it every day,” Raiza Martin, product manager at Google Labs, told The Independent.

Meanwhile, OpenAI has given us a voice mode to talk to, while still keeping the 128k simulation context window and, at legal risk, removed the hoarse Scarlett Johanson voice. The Gemini lineup was key to Google's superior performance. Share price has doubled since January 2023, reaching ATH on July 10, 2024.

Anthroponic: Ethics + Power = Big Money

Smaller than Google, but more important this year in terms of its role in the development of AI chatbots, is San Francisco-based Anthroponic. Founded by former OpenAI researchers, Anthroponic OpenAI has emerged as a formidable competitor in 2024, with explosive growth at the forefront and pockets of backers.

The rivalry between ChatGPT and Cloud, a large set of language models, is the equivalent of the Cold War in AI culture. When one company releases a feature, the other immediately hits. The two models are always competing for the top spot in the LLM arena and the community is always trying to decide which one is better.

Anthropic's revenue has grown more than 1,000% this year since the launch of Cloud 3.5 Sonnet in June, a major portion of which comes from third-party API users. Anthroponic Rise reflects the early days of OpenAI, but with a steeper trajectory.

And the investments will continue to flow. Amazon has invested $8 billion in the startup, while Google has agreed to invest up to $2 billion. These grants may challenge OpenAI's dominance and mirror the proxy war between tech giants and cloud computing providers, with Microsoft backing OpenAI to favor Azure and Amazon Cloud to favor AWS. Anthroponic and OpenAI mine the gold, while Microsoft and Amazon sell the shovels.

Anthropic's success is not just about the quality of the model. Overall, the company is pushing for a broader shift in the AI ​​landscape. As OpenAI chased consumer popularity, Anthropoc focused on unique models that prioritized security, betting that enterprise customers would pay a premium for targeted solutions.

Mistral AI: Making Europe relevant

In Europe, French startup Mistral AI raised eyebrows in June with $1 billion in funding and a $6 billion valuation. The open source models match the capabilities of GPT-4 at a fraction of the cost. To put this in perspective, this means “Mistral AI is worth €105m per employee” according to Trending Topics' calculations, making it the most valuable start-up in European history.

A recent update called “Le Chat” has positioned Mistral ChatGPT as a killer. It basically offers everything ChatGPT does – for free. It provides good results, captures code, generates images, supports agents and browses the web in real time.

And Mistral is planning to expand to the United States, opening a new office in the belly of the beast in Palo Alto, California.

China: Enforcing Sanctions Can't Stop Innovation.

Chinese companies aren't just copying – they're innovating. Even with massive U.S. sanctions to stifle innovation, companies like Baidu, Alibaba and Baichuan are developing models for emerging markets in Southeast Asia, the Middle East and Africa, backed by heavy government support. . They are building a parallel AI ecosystem that can rival anything coming out of San Francisco.

The state-backed telecom giant Huawei, for example, is developing its own operating system and embedding it into smartphones and home appliances, creating a fully functional ecosystem. Another Chinese model, Yi Lightning, defeated GPT-4o and Claude 3.5 Sonnet in the LLM arena, the new Deepseek came to compete with OpenAI's reasoning model o1, and Baidu's Ernie reached 100 million users in December 2023 and increased its user base to more than 200. million users in April 2024

According to the World Economic Forum, China's AI market is estimated to top $61 billion next year, with VCs investing more than $120 billion in AI ecosystems.

Meta: The best redemption arc in the tech industry

Believe it or not, Meta plays the role of the good guy in this story. While competitors like OpenAI and Anthropic battle for market share using expensive and locked-in models, Mark Zuckerberg's company took a different approach: acquiring the technology.

Llama 3.2, released in September, shows just how far Meta's open source strategy has come. The model will process both text and images, and is looking to work with US government agencies to use the model for everything from augmented reality applications to visual search engines and meta on national security applications.

And beyond governments and corporations, the Meta AI chatbot powered by Llama is a pretty promising competitor to ChatGPT. The chatbot, which has recently expanded to dozens of countries, can generate similar images with better quality than the Dall-e 3, and can even perform animations (which ChatGPT can't), search the web, perform coding tasks, and imagine scenes on location. Beyond that, Meta has other generative AI models for audio generation, video editing, segmentation, picture animation, and more.

With Lama 4 around the corner in 2025, it promises to add handling of text, audio and images, with meta bets opening and closing each time.

The numbers back up Zuckerberg's gamble. Meta AI, the company's answer to ChatGPT has amassed 500 million monthly users and is India's leader. Over a million advertisers jumped on the Meta AI bandwagon in September alone, churning out 15 million AI-generated ads. The company's revenue grew 18.9 percent year over year to $40.6 billion in Q3.

All this AI goodwill won't come cheap. Meta will pour $38 billion into capital spending this year, mostly on AI research and the hardware to run it. This includes powering its data centers with 350,000 of Nvidia's award-winning H100 AI chips by the end of the year. But Zuckerberg is playing the long game. Partnering with cloud giants AWS, Google Cloud and Microsoft Azure to host Llama models, it builds an ecosystem that shapes how Meta AI technology is deployed and who benefits from it.

Not bad for a company that lost 75% of its value when it started focusing on the “metaverse” and has grown 600% since turning to artificial intelligence.

Diversity is key.

Is OpenAI still king of the hill? Technically, maybe. At least the most recognizable and valuable in the picture – AI is a start. But the hill itself has changed. The competition isn't about raw power anymore – it's about trust, reach and real-world impact, and some of those areas, especially trust and security, are a little murky right now for Sam Altman's unicorn.

OpenAI's early leadership has evolved into a complex web of players, each carving out its own niche. Some focus on consumer applications, others on enterprise solutions, and a few brave people resist the basic research that AGI can unlock.

In the near future, Sam Altman seems very confident that OpenAI will reach AGI next year, which puts OpenAI at the top of the hill for a long, long time – and if that's an “if” – they'll succeed.

However, other highly respected and talented experts, such as Ian LeCun, Head of AI Research at Meta, believe that such an achievement could be achieved within 10 years or so. It depends on whether you decide to be an optimist or a pessimist.

The real winner of that 2024 AI race is the users. The competition drives innovation, but also forces companies to address concerns about security, privacy and accessibility. That's why the Amoday brothers left OpenAI to found Anthroponic and Ilya Sutskiver left to found Safe Superintendent. That's why Huawei developed its own mobile operating system and used local technology to build the best phones of the year, that's why developers come up with customized and better versions of the most popular AI models, and that's why AI is social as a technology. event in the last two years.

As these AI titans battle for supremacy, they are building tools that will change everything from how we do things to how we create and communicate.

And that ‘one AI to rule them all'? Maybe that was the wrong question all along. In the AI ​​landscape of 2025, diversity is bound to play a critical role. And that might be exactly what we need.