发帖
雷达卡
前言
TMT Predictions 2020: The canopy effect
2020年TMT预测：树冠效应
DELOITTE’S 2020 TMT Predictions report contains three overarching themes. First, individual technologies are no longer siloed, but are becoming ever more connected and interdependent—and their impact and value are increasing as a result. Second, most of the TMT industry’s money is coming from smartphones, computers, TVs, enterprise data centers and software, and IoT (we call these the “Big Five”). And third, many services and products that have been “just around the corner” for years are finally turning that corner in 2020.
德勤的2020年TMT预测报告包含三大主题。首先，单个技术不再孤立，而是变得越来越相互联系和相互依存，其影响和价值也因此而增加。其次，TMT行业的大部分资金来自智能手机、计算机、电视、企业数据中心和软件以及物联网（我们称之为“五大巨头”）。第三，许多多年来一直“近在咫尺”的服务和产品终于在2020年迎来了转折点。
Think of a forest…
想想森林…
In prior years of TMT Predictions, we have cross-referenced chapters a handful of times. This year, however, we have done so much more frequently. Consider that edge AI chips, private 5G, and robots are all interconnected,
在前几年的TMT预测中，我们有过几次交叉引用章节。然而，今年我们做得更频繁了。考虑到边缘人工智能芯片、专用5G和机器人都是相互连接的，
while ad-supported video and antenna TV are both affected by each other as well as by low-earth-orbit broadband satellites.
而支持广告的视频和天线电视既受近地轨道宽带卫星的影响，也受彼此的影响。
Why has this cross-linking ballooned in 2020?
为什么这种交联在2020年膨胀？
Think of a forest. In its youth, saplings grow meters apart from each other. Bacteria, fungi, insects, and animals coexist in a single tree, but the same organisms might not be present in the tree next to it. Each sapling is, to an
想想森林。年轻时，树苗彼此相隔几米长。细菌、真菌、昆虫和动物在同一棵树上共存，但同一生物可能不存在于它旁边的树上。每一棵小树都是
extent, an island with its own ecosystem. As the forest reaches maturity, while the trunks remain meters apart at ground level, 30 meters above the branches now all touch, creating a dense canopy that may be six meters thick.
一个拥有自己生态系统的岛屿。随着森林的成熟，树干在地面上相距数米，而树枝上方30米的地方都有了接触，形成了一个可能有6米厚的浓密树冠。
This single canopy, consisting of perhaps millions of trees, is now a unified ecosystem that may span thousands of kilometers.
这个由数百万棵树组成的单一的树冠，现在是一个可能跨越数千公里的统一生态系统。
A parallel phenomenon is underway in the TMT industry. Only 10 years ago, for instance, each kind of AI technology was its own “sapling”: Innovations in natural language processing did not lead to better visual recognition, for example. Then, new deep machine learning hardware began to accelerate all AI innovations at the same time, creating a “canopy” in which advances in one area were almost always matched by advances in the other (former) AI silos. Nor does the phenomenon stop there. Until recently, deep learning has been performed using chips that cost thousands of dollars and used thousands of watts, and that were hence largely restricted to data centers.
TMT行业也出现了类似的现象。例如，仅仅10年前，每种人工智能技术都是自己的“树苗”：例如，自然语言处理的创新并没有带来更好的视觉识别。然后，新的深度机器学习硬件开始同时加速所有人工智能创新，创造了一个“树冠”，其中一个领域的进步几乎总是与另一个（以前的）人工智能筒仓的进步相匹配。这种现象也不会就此停止。直到最近，人们才开始使用成本数千美元、耗电数千瓦的芯片来进行深度学习，因此这些芯片在很大程度上仅限于数据中心。
Within just the last two years, however, new edge AI chips that cost mere dollars and require only a few watts have made it possible to perform machine learning anywhere—further expanding the canopy, to stick with our metaphor.
然而，仅仅在过去两年内，新的edge人工智能芯片只需花费几美元，只需几瓦特，就可以在任何地方进行机器学习，进一步扩大我们的冠层。
Thanks to this development, even more data, algorithms, information, and solutions are flowing through all parts of the ecosystem, leading to ever faster and more useful AI for consumers and enterprises.
由于这一发展，更多的数据、算法、信息和解决方案正在流经生态系统的所有部分，从而为消费者和企业带来更快、更有用的人工智能。
DI_TMT-Prediction-2020.pdf
(9.03 MB, 需要: 20 个论坛币)
提升卡
置顶卡
沉默卡
变色卡
抢沙发
千斤顶
显身卡
京公网安备 11010802022788号
