NadiAIGlobal AI news. Concise. Clear. Easy to follow.
OpenAI says its custom inference chip, Jalapeño, achieves faster and more power‑efficient AI inference with higher throughput and lower latency for modern models. The publisher frames these as first results rather than independently verified performance benchmarks.
Benchmarking on SemiAnalysis’ InferenceX suite showed OpenAI’s Jalapeño chip delivered higher tokens per user and greater throughput per kilowatt than the current state-of-the-art. TechCrunch AI reported these test results without independent verification in the article metadata.
Anthropic is recruiting a chip design team to develop custom AI hardware. The company says it will co-design hardware and models to make its Claude technology run faster and more efficiently.
Alphabet is reportedly developing a new custom chip intended to make its Gemini models run more efficiently. TechCrunch reports the work aims to improve performance and efficiency for Google’s AI stack.
TechCrunch's Equity episode discusses whether Apple's lawsuit could affect OpenAI's reported hardware ambitions and plans to go public. The segment frames the lawsuit as a potential legal and strategic headache for OpenAI rather than a settled obstacle.
OpenAI has released a $230 illuminated keyboard intended to pair with its agentic coding app, Codex. The launch comes while the company is publicly engaged in a legal dispute with Apple over alleged hardware trade theft.
TechCrunch reports OpenAI’s first hardware may be a screenless, AI-guided smart speaker capable of movement. Details are limited and the report should be treated as tentative until confirmed by OpenAI.
Reports say Meta collaborated with Broadcom and TSMC on the design and manufacture of its Iris AI accelerator. Production could start as early as September, according to the report.
Meta plans to begin production of its new modular AI chips in September. The company says the design is modular to adapt as AI requirements evolve before and after deployment, according to TechCrunch AI.
Hugging Face Blog membentangkan perbandingan prestasi antara Habana Gaudi®2 dan Nvidia A100 80GB. Blog itu menyatakan Gaudi2 menunjukkan latihan dan inferens lebih pantas dalam beberapa konfigurasi ujian, tetapi pembaca digalakkan merujuk metodologi penuh dalam sumber.
Midjourney, yang terkenal dengan penjana imej AI, memperkenalkan produk perkakasan pertama mereka, The Midjourney Scanner, menurut CEO David Holz. Peranti itu digambarkan sebagai pengimbas ultrabunyi seluruh badan menggunakan cincin sensor untuk menangkap imej menegak, dan syarikat turut merancang sebuah spa di San Francisco.
