AI Wars: Nvidia, OpenAI Battle for Supremacy

OpenAI launched a new coding model, GPT-5.3-Codex-Spark, designed for near-instantaneous response times, marking its first major move beyond Nvidia's infrastructure, according to VentureBeat. The model, which runs on Cerebras Systems hardware, comes at a pivotal moment for OpenAI, amidst a frayed relationship with Nvidia, criticism over advertisements in ChatGPT, a new Pentagon contract, and internal organizational upheaval.

Nvidia researchers also announced a new technique, dynamic memory sparsification (DMS), that can reduce the memory costs of large language model reasoning by up to eight times, VentureBeat reported. This technique compresses the key value (KV) cache, the temporary memory LLMs generate, without degrading the model's intelligence. Experiments showed DMS enables LLMs to "think" longer and explore more solutions.

Meanwhile, other tech developments and controversies were also making headlines. Waymo, the self-driving developer, asked residents of Washington, DC, to pressure city officials to pass new regulations allowing its robotaxis to operate driver-free, according to Wired. The company has been pushing for these regulations for over a year, with plans to begin offering driverless rides in DC this year.

Wired also reported on a wave of unexplained bot traffic sweeping the web. One website owner, Alejandro Quintero, saw a sudden influx of traffic from China and Singapore, accounting for over half of his total visits in the past 12 months.

In other news, Wired covered ICE's plans to expand across the US, Palantir employees' ethical concerns, and the use of AI assistants. The publication discussed ICE's plans to expand to nearly every state in the US. The article also unpacked Palantir CEO Alex Karp's response to employee concerns about collaborating with ICE.