OpenAI moves to new chips as Samsung launches HBM4: a tech digest

# OpenAI moves to new chips as Samsung launches HBM4: where the AI market goes next

The AI technology landscape is going through a significant shift. OpenAI has introduced GPT-5.3-Codex-Spark, a model running on Cerebras chips. Samsung has started shipping HBM4 memory for accelerators. Startups such as MiniMax and ByteDance are showing alternatives in generative models. This is not just a routine update: it signals that Nvidia's monopoly on AI components is starting to crack.

OpenAI's move to Cerebras chips is symbolic. Years of refining Nvidia processors gave the company an unprecedented market advantage. Nvidia's A100, H100 and newer chips became the de facto standard for training large language models. But Cerebras, which specializes in processors for AI workloads, offers a different approach: an architecture optimized specifically for transformers and deep learning. OpenAI's decision to use these chips means it is actively diversifying its compute. The strategic move not only reduces dependence on a single supplier but also tells investors and competitors that viable alternatives have appeared.

At the same time, Samsung is expanding its presence in the critically important high-speed memory segment. HBM4 is a qualitative leap in high-bandwidth memory technology. HBM is essential for speeding up model training because it moves huge volumes of data quickly between compute elements. Samsung forecasts a threefold increase in sales of AI components, and this is not merely an optimistic projection. As more data centers train their own models, demand for high-quality memory will only grow. The company is betting that HBM4 standardization will be the next wave in the evolution of AI infrastructure.

Software competition is unfolding alongside the hardware innovation. MiniMax has released its M2.5 model, and ByteDance has presented the Seedance 2.0 video generator. These products challenge the dominance of OpenAI and Google in generative AI. Where model quality once depended directly on the amount of compute, architectural and methodological breakthroughs now let small companies compete with giants. This is the democratization of AI in action.

The financial sector is also reacting to the shifts. Elon Musk topped Forbes' list of innovators, ahead of Jensen Huang, the head of Nvidia. That is symbolic: despite Nvidia's dominance in hardware, public attention is shifting to the creators of services and platforms. SoftBank posted record profit thanks to investments in AI startups, and Lenovo reached an all-time revenue high by integrating neural networks into its products. This shows that value comes not only from the hardware itself but from its use in real-world solutions.

The picture emerging is of an AI ecosystem that is becoming more fragmented and competitive. New players are appearing in chips, memory and software. That is good for innovation, good for prices and good for the ecosystem as a whole. But it also means investors and developers must choose partners more carefully, because the technology landscape is becoming more complex and fast-changing. The future of the AI industry will not belong to a single player. It will belong to those who best integrate hardware and software into one system.