Big Tech’s arms race: how 2024 unfolded in AI

In 2024 artificial intelligence turned into an arms race. Almost every tech behemoth, a host of startups and firms across regions, from Europe to China, rushed to roll out their own AI products in an effort to chip away at OpenAI’s lead.

Here is a look at the notable launches and the most consequential events in the sector over the past 12 months.

Stock market

AI breakthroughs set the tone and drove US equities. Companies leaning hard into AI took turns at the top of the market-cap league table. 

In January Microsoft overtook Apple as the world’s most valuable firm, hitting a $2.9 trillion valuation. Its shares had surged over the prior year as investors rewarded its AI advances.

In June Nvidia for the first time became the most valuable company, powered by demand for its chips among large-language-model developers.

Hardware

Nvidia, the dominant supplier of AI processors, tightened its grip by unveiling the next generation of AI chips, Blackwell. 

The new parts deliver a hefty performance jump for neural-network workloads—20 petaflops versus 4 petaflops on the H100. CEO Jensen Huang said the added compute would let customers train larger, more complex models.

Two months later Nvidia announced a new training-processor architecture dubbed Rubin.

Demand for Nvidia’s processors remained intense. In January Meta said it would buy 350,000 Nvidia chips to expand its AI capacity, citing a “huge compute infrastructure” need on its AI roadmap.

In September xAI launched Colossus, billed as the world’s largest AI-training cluster, using 100,000 Nvidia H100 GPUs, according to Elon Musk. 

Rivals pushed their own hardware to claw market share from Nvidia. In April Intel introduced its next-gen AI chip, Gaudi 3, claiming over twice the energy efficiency of its predecessor and multiple configurations. 

In June Samsung outlined upcoming manufacturing improvements aimed at wooing AI-chip designers. Media also reported that OpenAI was working on its own solution.

OpenAI updates

OpenAI’s in-house chips are meant to train and run both legacy and freshly minted models, including those debuting in 2024.

The splashiest launch was Sora, a video generator. The product was announced in February, but the broader public gained access only in December. Users can create videos from text prompts, “animate” images, extend existing clips and fill in missing frames. 

It was a packed year for OpenAI. Beyond Sora, the company:

• unveiled a “more human-like” ChatGPT, GPT-4o, with the ability to process visual input;
• released GPT-4o mini, billed as “the most powerful and economically efficient small model available today”;
• developed the large language model o1, trained with reinforcement to tackle complex reasoning. The company says the model “can think”—it can form a long internal chain of thought while analysing a question;
• launched an expanded voice mode for ChatGPT;
• introduced Search in ChatGPT, which “allows you to get timely answers with links to relevant web sources”.

The firm also closed a $6.6 billion funding round at a $157 billion valuation, one of the largest venture deals on record.

Investment

Venture money did not flow only to OpenAI. Other multibillion-dollar deals included:

• Microsoft announced a long-term partnership with France’s Mistral AI, investing $2.1 billion;
• Amazon committed an additional $2.75 billion to Anthropic, bringing total backing to $4 billion;
• in April Elon Musk’s xAI raised $6 billion, and in December a further $6 billion at an implied valuation of $40–50 billion;
• Scale AI closed a $1 billion Series F round at a $13.8 billion valuation;
• Canada’s Cohere raised $500 million at a $5.5 billion valuation;
• Safe Superintelligence, founded by former OpenAI chief scientist Ilya Sutskever, raised $1 billion.

Medicine

Healthcare emerged as one of AI’s most promising use-cases in 2024. AI already helps researchers to:

• detect Alzheimer’s early;
• spot heart disease;
• interpret ambiguous medical images;
• identify 13 cancer types with 98.2% accuracy;
• effectively track Parkinson’s progression via hand movements;
• detect tuberculosis.

Not only benign aims

AI also found military uses. In January OpenAI quietly removed language from its use policy that banned “military” applications of ChatGPT. Later Meta and Anthropic took similar steps, opening AI technology to US government agencies and defence contractors and their allies.

Meanwhile, 200 staff at Google’s AI arm DeepMind voiced discontent over the firm’s military engagements, urging it to end contracts with defence and intelligence bodies. They fear the lab’s technologies are being sold to militaries involved in active conflicts.

China is also deploying AI for military purposes. A lab at the National University of Defense Technology’s College of Joint Operations in Shijiazhuang, Hebei province, began testing an AI “commander”. The country is running simulations of potential conflicts in the South China Sea, especially near Taiwan.

China

China worked to keep pace with America by building its own AI stack. The country far outstrips others in generative-AI patent filings, and its AI user base has reached 230 million. 

In June Alibaba announced its new model, Qwen2. It later began work on Tora, a Sora-like video generator, and also introduced the reasoning-focused QwQ-32B-Preview. 

Kuaishou moved to rival OpenAI in video generation by releasing the Kling model. Another startup, Zhipu, presented its own alternative, Ying.

Apple

In America, the AI pipeline was even richer, OpenAI aside. The headline act was Apple’s entrance.

Throughout the first half, rumours swirled about a coming release. In June Apple finally unveiled its sweeping initiative, Apple Intelligence. It includes:

• integrating ChatGPT into Siri and system-wide tools such as Writing Tools;
• enhanced photo search and clip creation; 
• transcription of email, messages and voice memos; 
• a focus mode to surface only important notifications; 
• the Writing Tools assistant;
• Genmoji, for on-the-fly emoji generation; 
• Image Playground, a platform for image creation; 
• Visual Intelligence, for analysing visual information; 
• Image Wand, which removes objects from photos;
• and numerous other features slated for release.

Billionaire Elon Musk threatened to ban Apple devices at his companies if ChatGPT from OpenAI is integrated into iOS, iPadOS and macOS.

Musk has a long-running feud with the startup he co-founded with Sam Altman. In February he sued OpenAI, accusing it of abandoning openness after releasing a closed-source AI product.

He argues that Sam Altman and OpenAI violated an agreement “to develop the technology for the benefit of humanity rather than for profit” dating back to the startup’s founding.

Musk later withdrew the suit but soon filed a new one with additional details. 

The entrepreneur also faced a lawsuit from Tesla shareholders over xAI. They say the startup competes with Tesla in AI, “siphoning the automaker’s talent and resources”. At least 11 Tesla employees moved to xAI, and Tesla provided the startup access to AI-related data, the filing claims.

AI products from others

Other tech giants kept up a brisk cadence of AI releases. Standouts included:

• integrating generative AI into Google Search, Chrome, Gmail, new Pixel phones and other services, and releasing its “most powerful to date” AI model, Gemini 2.0;
• Meta’s launch of the video generator Movie Gen, the Llama 3.1 model family followed by Llama 3.2, user-created AI characters in Instagram and the Meta AI assistant across WhatsApp, Instagram, Facebook and Messenger;
• xAI’s April update to Grok-1.5 and August launch of Grok-2 with image generation;
• Microsoft’s revamped Copilot, adding voice and on-screen image understanding.

In Europe, Mistral remained the standard-bearer. In July it released its flagship Large 2 model, and in September debuted its first multimodal network, able to process images and text. 

The crypto–AI symbiosis

2024 was a breakthrough year for the fusion of cryptocurrencies and AI. Some venture capitalists voiced scepticism, while others highlighted the benefits. 

In January Ethereum co-founder Vitalik Buterin outlined four promising overlaps: countering AI centralisation, ensuring transparency, and improving data storage and security.

He later considered using AI for formal code verification and bug-finding.

Meanwhile Binance said it used AI to thwart fraud. The exchange prevented more than $2.4 billion in user-fund thefts in 2024, thanks to the technology. In August Binance Labs investment chief Max Konilio explained the firm’s AI push, calling AI and blockchain “a powerful combination”. AI, he argued, improves blockchain user experience and decentralised ecosystems, while blockchains provide authenticity, foster open-source collaboration and support AI compute needs.

In November former Binance CEO Changpeng Zhao (CZ) urged building AI products with blockchain. In his view, the two are well suited to work together.

Coinbase also dove into AI—over the summer it built and deployed a machine-learning model to predict traffic spikes and auto-scale databases to reduce downtime and improve performance. Later the exchange introduced a service for building AI agents for crypto wallets, while Tether rolled out a tool for privacy-focused applications.

In June Bitwise senior crypto research analyst Juan Leon predicted that by 2030 AI and digital assets would add $20 trillion to global GDP. He flagged partnerships between miners and AI firms as one possible avenue of collaboration.   

The AI-token boom

Rapid progress in AI, memecoins and crypto—and the prospect of merging them—spawned a new class of digital assets: AI tokens. 

The first boom came in the year’s first half amid a wider crypto rally. In February the AI-token segment surged, pushing aggregate market cap above $15 billion, likely on excitement over OpenAI’s Sora video model. Prices jumped again after Nvidia’s earnings.

The market then corrected, and eye-catching headlines returned in autumn. In October a new memecoin, Goatseus Maximus (GOAT), soared nearly 2,000% in two days on rumours of support from the AI bot Truth Terminal. The project ai16z also drew attention, later reaching a $100 million market cap.

AI agents

A nascent narrative is AI agents—systems that can act online without human input, such as booking flights or hotels. Some may issue their own tokens, a trend likely to catch on in crypto circles.

Recent launches include:

• Anthropic’s update—its startup released Claude 3.5 Sonnet with the ability to use a computer like a human: moving the cursor, clicking buttons and typing;
• OpenAI’s work—the company is preparing an agent codenamed “Operator”, capable of using a computer to write code or book trips on a user’s behalf;
• Google’s tool—the company’s AI unit presented its first AI agent able to act on the web autonomously.

Microsoft has also signalled plans to launch AI agents. 

Regulation

As AI raced ahead, regulators worked up legal frameworks.

In March the European Parliament approved the world’s first comprehensive AI rulebook. It took effect on August 1 and drew a wave of criticism. 

In America, California’s SB 1047—aimed at regulating AI—was debated. It won support from Elon Musk and Vitalik Buterin but never became law after the governor declined to sign it. 

In December US president-elect Donald Trump nominated David Sacks as “czar” for AI and digital assets, naming them priorities of his administration.

Conclusions

Above all, 2024 showed that everyone is betting on AI: startups, tech giants, venture capitalists, crypto users and even governments.

It is increasingly clear that AI is a revolutionary technology on a par with the smartphone—and it is here to stay. 

Stay with ForkLog for hot AI-sector headlines in 2025.