谷歌云在怕些啥？

谷歌云发生了什么？ 今天，乔恩、凯文和阿米尔有一个关于谷歌美国最高销售主管克尔斯滕克里普豪斯离职的高管改组的重要故事。 你可能还记得，谷歌云经过 10 年的尝试，在业务上仍远远落后于亚马逊和微软。

谷歌不太喜欢这个故事，并开始告诉人们这是错误的。 当我们追问原因时，该公司表示这是不准确的，因为接替 Kliphouse 的人同时也在负责客户支持。 好的。 但是 Kliphouse 出局了。 新人 Adaire Fox-Martin 负责监督销售（以及客户支持）。 故事结局。

这是一个经典的误导工具。 虽然我通常不会公开写下所有混淆或说谎的公司企图，但我今天这样做是因为这很重要，而且这一事件很能说明问题。

谷歌正在审视其未来 30 年的商业模式。 其成熟的搜索广告业务不会成为其未来的增长引擎。 从理论上讲，云会是，但该部门即将到来的销售情况看起来很艰难，因为您可能还记得我们第一次报道谷歌云的客户合同疲软时的情形。 与此同时，亚马逊未来来自云计算的预期收入继续飙升。

难怪谷歌很敏感。 考虑到 Jon 昨天在 Alphabet 的两个部门——机器人部门 Intrinsic 和生命科学项目 Verily 爆出的裁员消息——脸皮薄的原因变得更加清晰。

一段时间以来，我觉得谷歌必须有更多的底牌。 长期以来，这家公司一直是杰出工程师和研究人员的家园，尤其是在人工智能领域，该公司似乎处于领先地位。 我一直觉得它一定有一些秘密的人工智能，可以在云端轻松与亚马逊和微软竞争。 如果它真的让他们更成功，哪个企业不会报名参加呢？

虽然谷歌为企业宣传其机器学习工具，但我们仍在等待能够赢得销售的重大成果。 与此同时，众所周知，微软和 OpenAI 正在争先恐后。

作为一个长期的谷歌观察者，老实说我有点困惑。 而像他们今天针对我们的竞选这样的恶作剧肯定让我怀疑：他们真的担心那边吗？

似乎是这样。

Rent the Runway 转向了亚马逊云

首先是 Peloton，现在是 Rent the Runway：亚马逊正在成为苦苦挣扎的零售商出售商品并快速赚取现金的渠道。 这家服装租赁服务公司周四宣布，将在亚马逊上开设一个店面，出售人们租用过的二手服装和 The RTR Designer Collective 独家品牌的新品。 Rent the Runway 将这笔交易描述为有助于提高其服务的知名度。 但另一种解释是，该公司正在转向成为亚马逊的供应商。

好吧，不完全是。 Rent the Runway 仍在出租其服装。 但亚马逊的两笔交易已经证明 Rent the Runway 是有利可图的，该公司在 10 月季度的利润达 460 万美元，与“试点计划”和与一家身份不明的零售商的清算合作伙伴关系有关，我们现在知道是亚马逊。 这 460 万美元占本季度好于预期的“调整后 Ebitda”的 70%，Rent the Runway 青睐的盈利指标（正如我们之前指出的那样，这是一个可疑的指标）。 似乎亚马逊去年从 Rent the Runway 购买了这些商品，比发布早了几个月。

投资者面临的问题是，与亚马逊的合作是否会继续尚不清楚。 在 12 月讨论 10 月季度收益的电话会议上，Rent the Runway 的高管告诫分析师不要假设当时神秘的交易会重演对利润的影响。 换句话说，亚马逊购买了一定数量的商品，可能不会再购买了。 如果店面证明很受欢迎（正如 The Information 的 Ann Gehan 最近预测的那样），也许亚马逊可以直接收购 Rent the Runway 本身。 至少，投资者并没有留下太深刻的印象。 Rent the Runway 股票周四小幅下跌。—Martin Peers

人工智能的未来一年
周四，The Information 的订阅者 Hugging Face 首席执行官 Clément Delangue 和 Character.AI 首席执行官 Noam Shazeer（谷歌人工智能实验室的前主要负责人）以及我们的一些记者和编辑讨论了人工智能的未来一年，他们一直在谈论关于 OpenAI 和相关公司几个月。

一个突出的收获是 Delangue 的观点，即初创公司可以从大量的 AI 工程人才中吸取经验，以更好地与拥有更多资源的科技巨头竞争。 与此同时，Delangue 还表示，许多此类初创公司仍然依赖大型云提供商，这些提供商通过向他们发放信贷来有效地赞助他们的运营。 包括谷歌和微软在内的云提供商将这种做法视为与前沿人工智能开发人员保持密切联系的一种方式。 “如果这种理论发生变化，云提供商开始想从中赚钱，那么很多公司……可能会陷入困境。”他说。

尽管关于 OpenAI 的 ChatGPT 等新工具的威胁的讨论已经持续了数周——它会根据简单的文本提示给出类似人类的答案——对谷歌搜索构成威胁，但 Shazeer 并不买账。 他说，谷歌拥有人才、专业知识、十亿用户产品和大量私人数据，可以从实施更新的人工智能技术中受益。 这就是谷歌十年前所做的事情，当时深度学习变得可行，可以根据每个用户的活动改进其广告定位和 YouTube 等个性化服务。

电话会议的参与者谈到了另一个问题。 机器学习模型越来越大，使用它们自动执行复杂任务并不便宜。 然而，我们正处于经济低迷时期，公司（即软件购买者）正在裁员并放慢或减少他们对软件的购买。 正如 The Information 报道的那样，这些因素已经使 AI 软件在企业界的采用变得复杂和缓慢。

What is going on at Google Cloud? Today, Jon, Kevin and Amir had an important story about an executive shakeup involving the departure of Google’s top sales executive in the U.S., Kirsten Kliphouse. You may recall that Google Cloud, after 10 years of trying, is still a distant second to Amazon and Microsoft in the business. 

Google didn’t much like the story and started telling people it was wrong. When we pushed it on why, the company said it was inaccurate because the person replacing Kliphouse was also looking after customer support. OK. But Kliphouse is out. And the new person—Adaire Fox-Martin—is overseeing sales (as well as customer support). End of story.

This is such a classic misdirection tool. And while I don’t usually write publicly about all the obfuscation or lying companies attempt, I am doing so today because it is important and this incident is telling. 

Google is looking at its business model for the next 30 years. Its maturing search ads business won’t be its growth engine in the future. In theory, cloud will be, but the unit’s upcoming sales picture looks tough, as you may recall from when we first reported that Google Cloud’s customer contracts have been anemic. Meanwhile, Amazon’s future expected revenue from cloud continued to soar.

No wonder Google is sensitive. When you consider the layoff news Jon broke yesterday at two Alphabet units—robotics unit Intrinsic and life sciences project Verily—the reasons for that thin skin become clearer.

For a while now, I’ve felt that Google must have more cards up its sleeve. Long a home for preeminent engineers and researchers, especially in artificial intelligence, the company seems so well placed to lead. I’ve always felt it must have some secret AI that could easily compete with Amazon and Microsoft in the cloud. What business wouldn’t sign up for that, if it truly made them more successful?

And while Google plays up its machine-learning tools for businesses, we’re still waiting for the big stuff that wins sales. Meanwhile, as we all know, Microsoft and OpenAI are chomping at the bit. 

As a longtime Google watcher, I’m honestly a little confused. And shenanigans like their campaign against us today definitely make me wonder: Are they really worried over there?

Seems so.

Rent the Runway’s Amazon Pivot

First Peloton, now Rent the Runway: Amazon is becoming the outlet for struggling retailers to offload merchandise and generate some quick cash. The clothing rental service announced Thursday it will launch a storefront on Amazon to sell both used clothing—which people have rented—and new items from The RTR Designer Collective’s exclusive brands. Rent the Runway portrayed the deal as helping raise awareness of its service. But another interpretation is that the company is pivoting to become a supplier to Amazon.

Well, not quite. Rent the Runway is still renting out its clothes. But the two Amazon deals together have already proved profitable for Rent the Runway, which booked profits amounting to $4.6 million in the October quarter related to a “pilot program” and liquidation partnership with an unidentified retailer we now know is Amazon. That $4.6 million amounted to 70% of the quarter’s better than expected “adjusted Ebitda,” Rent the Runway’s favored metric of profitability (and a dubious one, as we’ve pointed out previously). It seems Amazon bought the merchandise from Rent the Runway last year, months ahead of the launch. 

The problem for investors is that it’s not clear the venture with Amazon will continue. On a conference call in December to discuss October-quarter earnings, Rent the Runway executives cautioned analysts not to assume the profit impact from what was then a mysterious deal would be repeated. In other words, Amazon bought a certain amount of merchandise and may not buy any more. Perhaps Amazon could simply buy Rent the Runway itself if the storefront proves popular (as The Information’s Ann Gehan predicted recently). Investors, at least, weren’t too impressed. Rent the Runway stock dipped slightly Thursday.—Martin Peers

The Year Ahead in AI

Subscribers to The Information on Thursday chatted about the year ahead in AI with Hugging Face CEO Clément Delangue and Character.AI CEO Noam Shazeer—a former key leader at Google’s AI lab—along with some of our reporters and editors who have been breaking news about OpenAI and related companies for months.

One takeaway that stood out was Delangue’s point that startups can draw from a wealth of AI engineering talent to better compete with tech giants that have more resources. At the same time, Delangue also said many such startups still rely on large cloud providers that effectively sponsor their operations by issuing credits to them. The cloud providers, including Google and Microsoft, view the practice as a way to stay close to bleeding-edge AI developers. “If this theory changes and the cloud providers start to want to make money out of this, a lot of companies…could be in trouble,” he said.

Despite weeks of chatter about the threat new tools such as OpenAI’s ChatGPT—which gives humanlike answers in response to simple text prompts—pose for Google Search, Shazeer doesn’t buy it. Google has the talent, know-how, billion-user products and oodles of private data to benefit from implementing newer AI tech, he said. That’s what Google did a decade ago when deep learning became viable for improving its ad targeting and personalizing services like YouTube based on each user’s activities. 

Participants on the call talked about another problem. Machine-learning models are getting bigger and using them to automate complex tasks isn’t cheap. And yet we’re in the middle of an economic downturn in which corporations (that is, software buyers) are cutting staff and slowing or reducing their purchases of software. Those factors have already complicated and slowed the adoption of AI software in the corporate world