AI芯片行业正迎来一股初创公司热潮，它们专注于与模型训练截然不同的多样化工作负载——推理。Groq、Cerebras Systems、SambaNova和Lumai等公司正在开发专用硬件，包括光学加速器，以满足推理任务的多样化需求。这一转变为了这些初创公司提供了与英伟达等成熟企业竞争的机会，因为AWS和谷歌等主要云服务提供商也在探索分解式计算平台和定制加速器。 AI
影响 新兴的推理硬件初创公司和分解式计算平台可能为占主导地位的GPU供应商提供替代方案，从而可能降低成本并提高专业化水平。
排序理由 多家初创公司和云服务提供商专注于推理硬件开发，表明AI基础设施格局发生了重大转变。
AI 生成摘要 · Google Gemini · 来自 5 个来源。 我们如何撰写摘要 →
<h4>In a disaggregated AI world, Nvidia can be both a friend and an enemy</h4> <p>AI adoption is reaching an inflection point as the focus shifts from training new models to serving them. For the AI startups vying for a slice of Nvidia's pie, it's now or never.…</p>
AI-chip startups tar upp kampen mot Nvidia i inferensens nya era https:// redaktionen.net/artikel/832 # ai # svtech
Google controls ~25% of global AI compute — 3.8 million TPUs and 1.3 million GPUs. When a single actor holds that much of the world's AI infrastructure, the security and resilience questions go way beyond a single CVE. Concentration of compute *is* a threat model. A fascinating a…
Google controls ~25% of global AI compute, with ~3.8M TPUs and 1.3M GPUs https://www. ft.com/content/2429f0f0-b685-4 747-b425-bf8001a2e94c # HackerNews # Google # AI # compute # TPUs # GPUs # technology
Google controls ~25% of global AI compute, with ~3.8M TPUs and 1.3M GPUs https://www.ft.com/content/2429f0f0-b685-4747-b425-bf8001a2e94c # HackerNews # Tech # AI