按照这个问题的清单来然后我们的第一个问题就是因为前几天呢OpenAI也开了这个发布会那么这次它发布的这个GDPCO的话就是您说您认为它最大的看点是在哪里然后以及就是我们从用户角度来看的话其实它的这个交互体验的一个提升是非常明显的当然这个肯定是它延迟的一个降低所以您能否详细给我们介绍一下就是在这块的OpenAI它使用的一个技术 OK 我其实觉得体验降低那个事情倒还好因为它基本上是一个纯工程问题另一个很大的亮点其实你看它叫GBT-4O其实可能意味着至少这个模型没有从预训练这个阶段推导重新来一遍所以它单纯通过后面的fine tuning的方式可以实现多模态至少OpenAI自己宣称这个是一个原生多模态的模型 所以这个其实是一个非常神奇的一点就是怎么做到的因为至少从Google来说基本上是每轮都从预训链开始重新跑一遍的Five Tuning可能很难达到这个效果呃这个是一点然后对于比方说呃这个体验的提升其实从呃GBT-4刚发布到今天可能推理的速度吧呃 我觉得可能提升了至少十倍应该十倍不止因为你比方说去年GBT4刚发布的时候可能推理的速度大概是十个token一秒而且有可能随时当机对吧服务不稳定但是像今天比方说OpenAI自 ...

太平洋时间5月14日10点，GoogIe召开2024 IO大会 内容来自Google官方发布会，视频链接：https://io.google/2024/intl/zh/ Gemini大模型 现有成果： 超过 150 万开发人员在使用 Gemini 的API 我们拥有20 亿用户的产品都使用了 Gemini。 ...

Googlel/o开发 116 能股份 海外，OpenA 行业机构关注度 值得买 +2.4% 昆仑万维 +0.46% ...

Googlel/o开发 116 能股份 海外，OpenA 行业机构关注度 值得买 +2.4% 昆仑万维 +0.46% ...

Investment Rating - The report highlights 16 Buy-rated stocks across sectors exposed to the US data center power demand surge [7][8][49]. Core Insights - The report forecasts that global data center power demand will more than double by 2030, driven by AI and broader demand, leading to a 2.4% CAGR in US electricity demand through the end of the decade [2][6][49]. - Data centers are expected to comprise approximately 8% of total US power demand by 2030, up from about 3% currently, with a projected 160% increase in data center power demand from 2023 to 2030 [11][49]. - The report estimates that about $50 billion in capital investment will be required for new power generation capacity in the US to support this growth, with a 60/40 split between gas and renewables [2][11][49]. Summary by Sections Data Center Power Demand - Data center power demand is projected to grow at a 15% CAGR from 2023 to 2030, contributing approximately 90 basis points to the overall US power demand CAGR [6][49]. - The report anticipates that data centers will account for 8% of total US power demand by 2030, translating to a 16% CAGR for data centers from 2023 to 2030 [11][49]. Investment Opportunities - The report identifies underappreciated equity investment opportunities in utilities, renewable generation, and industrials that will support the anticipated growth in data center power demand [2][7]. - It emphasizes that investor interest in the data center growth is not new, but differentiation among enablers and beneficiaries will be crucial for future investment strategies [8][49]. Infrastructure and Capacity Needs - An estimated 47 GW of incremental power generation capacity will be required to support US data center power demand growth cumulatively through 2030 [6][11]. - The report discusses potential bottlenecks in generation and transmission, highlighting the importance of expediting the permitting process for new projects to alleviate constraints [6][20]. AI's Role in Power Demand - AI is expected to represent about 20% of overall data center power demand by 2030, with a projected increase of around 200 TWh in power demand from AI between 2024 and 2030 [22][29]. - The report notes that new AI innovations are increasing computing speed while also raising power consumption per server, indicating a complex relationship between demand and efficiency [31][38].