[Digital Today reporter Chi-gyu Hwang] Big Tech companies in the AI ecosystem appear to be gaining momentum in acquiring startups. Big Tech-led M&A cases are continuing across AI chips, large language models (LLMs) and AI applications.

Nvidia and Meta have drawn attention recently. Nvidia acquired assets of AI chip startup Groq for $20 billion. Through the Groq acquisition, Nvidia is expected to accelerate its push to expand influence in the fast-growing inference AI market.

Nvidia's move to effectively acquire Groq, an AI chip startup focused on AI inference, is unlikely to affect only the AI chip market. Many analyses say it is also an issue worth watching in areas such as high-performance memory like HBM and cooling technology, which has emerged as a variable in the AI infrastructure space.

Nvidia is also in advanced talks to acquire LLM startup AI21 Labs for up to $2 to $3 billion, reports say. AI21 Labs developed an open-source LLM called Jamba. Jamba processes data 2.5 times faster than existing algorithms, the result of optimising memory use by combining a transformer model with a neural network architecture called Mamba. Earlier, Nvidia acquired SchedMD, which developed the open-source cluster management software Slurm, in early December.

Meta acquired Singapore AI startup Manus. Manus has drawn attention as an AI agent that performs various tasks using a web browser without human intervention.

SoftBank will acquire DigitalBridge Group for $4.04 billion to expand AI infrastructure. DigitalBridge is a New York Stock Exchange-listed company that manages $108 billion in assets and owns data centre builder Vantage Data Centers and edge data centre operator AtlasEdge, among others.

Moves by domestic and overseas tech companies seeking to develop or use AI directly also continued this week. OpenAI is focusing on advertising as one of ChatGPT's revenue models, but it has been unclear how it would deliver ads, and it appears to be exploring various tactics internally. OpenAI employees are also discussing, as one of several options, providing sponsor information first in responses that ChatGPT delivers. Japan's SoftBank has executed the full $40 billion it pledged to invest in OpenAI.

Apple delayed the launch of an artificial intelligence-based Siri and took a major hit in the 2025 AI race. But it appears to be seeking a turnaround in 2026 by making sweeping improvements in AI technology.

Naver Cloud signed a memorandum of understanding with JB Financial Group (Gwangju Bank, Jeonbuk Bank, JB Woori Capital) to innovate financial services based on AI technology. Naver Cloud and the three JB Financial Group units plan to cooperate on identifying digital finance innovation tasks based on AI technologies such as HyperCLOVA X and AICC, advancing the latest IT technologies such as AI, big data and cloud and business cooperation, and joint AI research and development of specialised models for the financial sector.

JobKorea will fully launch a strategy to shift to an "AI-native recruitment platform" starting with its 30th anniversary this year. Through advancing its in-house AI solutions and commercialising an AI career agent, it aims to present a new recruitment paradigm of "probability-of-success-centred hiring". It also plans to accelerate its shift to an "AI-native organisation" by embedding AI across services, branding and overall organisational operations, following its acquisition of JobPlanet.

Generative AI startup Scatter Lab upgraded the language model for its in-house AI fiction platform Zeta and strengthened the user experience so longer stories can maintain a stable flow.

The first evaluation results for 5 elite teams participating in a government-led project to develop an independent foundation model will be released in January. As a result, 1 company among the 5 selected elite teams will be dropped. Earlier, the Ministry of Science and ICT and the National IT Industry Promotion Agency held the first presentation for the "independent AI foundation model" project in late December.