How DeepSeek and Next-generation aI Agents might Erode Value Of Language Models

In the internet revolution, we’re moving from developing websites as the main company to really companies – so, the Airbnb of AI, the Stripe of AI,” he added. “They are not about the design. They are about the system and how you make the design helpful for tasks.”

The arrival of DeepSeek’s R1 design recently is viewed by some tech CEOs as an additional indicator LLMs are ending up being increasingly commoditized.

R1 includes some key elements that distinguish it from other designs on the market. For example, the design incorporates a “combined accuracy” structure that uses a combination of full-precision 32-bit floating point (FP32) numbers and low-precision 8-bit drifting point (FP8) numbers.

The latter is quicker to process but can be less accurate. However, rather than depend on one or the other, DeepSeek uses FP8 for the majority of computations and changes to FP32 for certain jobs where a greater degree of precision is needed.

“As AI gets more effective and available, we will see its usage skyrocket, turning it into a product we just can’t get enough of,” Microsoft CEO Satya Nadella composed on social networks platform X, in an evident recommendation to DeepSeek.

Meanwhile, Matt Calkins, CEO of U.S. software firm Appian, told CNBC that DeepSeek’s success just shows that AI designs are going to end up being more of a product in the future.

“In my viewpoint, we’re visiting a commoditization of AI. Many business will accomplish competitive AI, and a lack of differentiation will be bad for high-spending first-movers,” Calkins stated by means of e-mail.