How Microsoft’s CFO Calls the Shots on AI Spending

[00:00:00] Exchanges. The Goldman Sachs podcast featuring exchanges on the forces driving the markets and the economy. Exchanges between the leading minds at Goldman Sachs. New episodes every week. Listen now.

[00:00:18] Welcome to Tech News Briefing. It's Friday, November 8th. I'm Cordelia James for The Wall Street Journal.

[00:00:25] Fake online reviews can be tricky to spot, but they don't have to be. We walk you through what to look out for so you don't get duped.

[00:00:33] And then, the woman behind Microsoft's multi-billion dollar artificial intelligence budget is Amy Hood, the company's chief financial officer.

[00:00:42] WSJ reporter Tom Doton tells us how she is steering the company's massive spending on AI.

[00:00:53] First up, when checking reviews online, it can be hard to know what to believe.

[00:00:58] According to some estimates, 16 to 40 percent of reviews are phony.

[00:01:02] To curb this spread, the Federal Trade Commission created a rule that finds companies that pay for fake reviews.

[00:01:10] Heidi Mitchell wrote about this for The Wall Street Journal, and she joins us now with more.

[00:01:15] Heidi, what are the websites that host reviews, like Google and Yelp, doing about this problem?

[00:01:20] Everyone that I spoke with, they're really working hard to solve for it.

[00:01:25] So they have software that will read everything before it goes through.

[00:01:29] A lot of them have humans also on staff whose job is to read ones that the software flags.

[00:01:36] They will ban people from sites.

[00:01:38] They recognize patterns and prevent those emails or specific users from posting.

[00:01:45] Okay, so what's the first thing people should do to spot a fake review?

[00:01:51] The first thing you should know is that you're really bad at seeing if it's a real review or a fake review.

[00:01:56] Humans are like 50 percent right.

[00:01:58] It's just a coin toss.

[00:02:00] So you really should go in reading with just that mindset.

[00:02:03] But the one thing that researchers say you can look for are pictures of the reviewer.

[00:02:07] One researcher that I spoke with said that 46.9 percent of the reviews that they deemed to be fake did not have a profile picture of the reviewers.

[00:02:18] You also wrote that writers of these reviews might use the word we instead of I.

[00:02:23] Why is that?

[00:02:24] Yeah, this one I thought was really interesting.

[00:02:26] So one study showed that fake reviewers, these are mostly humans that are writing these,

[00:02:32] that they kind of want to have a psychological space between them and their reader.

[00:02:36] They know what they're doing is wrong or at least not real.

[00:02:40] And so they don't use the word I.

[00:02:43] They may include like a pretend family or when we were at the restaurant and that kind of gives them some mental distance between it.

[00:02:53] It's kind of a guilt check in their own conscience.

[00:02:56] Wow.

[00:02:57] And what do we know about the timing of fake reviews?

[00:02:59] If you look at the timestamps on them and if they came in kind of a cluster,

[00:03:04] if you see like 50 reviews that were all within a few minutes of each other,

[00:03:08] that's a pretty good sign that somebody was paying for these because they tend to stream into the platforms.

[00:03:14] And then when the money runs out or they've hit their quota, they just shut off the spigot.

[00:03:19] Let's talk more about what's in these fake reviews.

[00:03:22] What do they look like?

[00:03:23] So they tend to be pretty short because even if you're a human,

[00:03:27] you're not going to spend that much time writing a very long poetic response.

[00:03:32] And they're paid for the volume of reviews, not the length typically.

[00:03:35] So they'll be short.

[00:03:37] They are typically flooded with lots of extremes.

[00:03:40] Like it's awesome!

[00:03:42] It's awful!

[00:03:44] One thing that you should think about is that fake reviewers are going to be pretty much at the extreme.

[00:03:50] So at the five-star and the one-star, it's better to just look at the healthy middle.

[00:03:55] Two- to four-star reviews are likely to be more meaningful and real than the one- and five-stars,

[00:04:02] which may be fake because they're trying to boost or lower the overall score.

[00:04:07] That was WSJ contributor Heidi Mitchell.

[00:04:10] Coming up, Microsoft CFO Amy Hood is at the center of the company's multi-billion dollar bet on AI.

[00:04:17] Who she is and why she supports this huge investment after the break.

[00:04:30] Dieser komplexe Finanzierungstalk ist ganz schön anstrengend.

[00:04:33] Ob ich mein Depot jemals angelegt kriege?

[00:04:35] Aber du hast doch schon ein Depot.

[00:04:37] Äh, nee.

[00:04:38] Doch, du hast das Vodafone Gigadepot.

[00:04:40] Ach, stimmt!

[00:04:41] Und da habe ich ja selbst in der Hand, wie groß mein Depot ist.

[00:04:44] Jetzt mit dem Vodafone Gigadepot und verbrauchtes Datenvolumen in den nächsten Monat mitnehmen.

[00:04:48] Go on im zuverlässigen 5G-Netz von Vodafone.

[00:04:51] Vodafone. Vodafone. Together we can.

[00:05:00] Satya Nadella might be Microsoft's CEO, but projects don't get off the ground without input from Amy Hood first.

[00:05:07] During her 11-year tenure as Microsoft's chief financial officer,

[00:05:11] Hood has been at the center of some of the biggest decisions that have made Microsoft the most acquisitive company in the tech industry.

[00:05:18] Now, she's steering tens of billions in new spending on AI.

[00:05:23] WSJ reporter Tom Doton joins us now with more.

[00:05:27] So Tom, you write that Amy Hood is not a technologist and she doesn't have a deep background in accounting.

[00:05:33] So how did she build trust with her colleagues and rise to the role of CFO?

[00:05:37] She came on working for their investor relations team.

[00:05:43] So it's somewhat adjacent to the work that she had been doing for Goldman Sachs.

[00:05:47] And she quickly proved in the company to be a very sharp operator, someone who really understood the numbers of how Microsoft works.

[00:05:54] And she rose fairly quickly within the company, starting from the investor relations team, then going on to eventually being the chief financial officer divisional CFO within the office group, which is an important business there.

[00:06:08] And then at the tail end of Steve Ballmer, who was the previous CEO of Microsoft's regime, she was appointed as the company CFO.

[00:06:16] And she's someone who people saw as very sharp, very willing to get into the details of a business.

[00:06:25] And Microsoft is historically a fairly conservative financial company.

[00:06:29] They don't like to spend money on things that don't have clear business returns.

[00:06:34] And she's someone who kind of was able to understand businesses well enough that she could lead those kinds of business moves.

[00:06:41] We're going to get into those business moves in just a second.

[00:06:43] But first, can you tell me a little bit about what her relationship with Microsoft CEO Satya Nadella is like?

[00:06:50] She actually met Nadella on her way up.

[00:06:52] He was a top executive within the servers and tools division, which ended up becoming Azure, which is like their really important cloud computing business.

[00:07:00] And when he became CEO, she stayed on as CFO.

[00:07:04] And so they have a history together.

[00:07:06] And talking to sources in and around Microsoft, they have a very starry-eyed dreamer and grounded realist dynamic.

[00:07:16] So Satya is someone who fashions himself as something of a visionary who tries to look at what the next trend is going to be and make sure Microsoft is ready for that.

[00:07:24] And she's someone who obviously is there to make sure that those visions can be realized financially, but also have a little bit more grounding in what the business will allow for, what investors will allow for, what the board is down for.

[00:07:37] Let's talk about some of those key financial decisions that she's making to keep things grounded there.

[00:07:42] What were some of those decisions and how have they helped shape the Microsoft that we know today?

[00:07:47] I was really interested in, for this story, AI and their $64 billion capital expenditure in AI.

[00:07:56] But obviously getting there took some time and she's made some interesting decisions along the way.

[00:08:02] There was an effort a couple of years ago to acquire this company called Truecaller, which was like a phone app for screening out spam phone calls.

[00:08:11] The Skype team wanted to acquire them and she pushed back on that saying Microsoft really shouldn't be acquiring consumer apps, that the marketplace wasn't large enough.

[00:08:19] A couple of years later, when they were in the midst of the explosion of personal assistants, digital assistants, Amazon was hiring up like crazy on their Alexa team.

[00:08:30] And people at Microsoft were seeing that saying, oh, why aren't we doing that?

[00:08:34] This is like the next big frontier.

[00:08:35] Why doesn't Microsoft just also go as extremely into our personal assistant?

[00:08:39] And she was like, nope, there's no market for that.

[00:08:42] It's not worth us investing that much money in it.

[00:08:44] And you could argue that she's really been proved right because of all of that.

[00:08:48] And then as a counter to some of that, when Microsoft was in negotiations to acquire LinkedIn in 2016, they really wanted it.

[00:08:57] Satya, this was like an absolute must have for him.

[00:09:00] And they were in a bidding war with Salesforce.

[00:09:02] And the price kept going up and up because this was a hot asset.

[00:09:06] And Amy went to Satya in a meeting in the midst of the negotiation and basically said, whatever price you say you want to get this done at, let me know.

[00:09:14] And I will adjust the business case to the board to make sure that we get approval for it.

[00:09:19] So in one sense, she is willing to push back on what she sees as not well thought out or well grounded acquisitions.

[00:09:26] But if it does make sense, she will be in favor of spending maybe more than an asset may initially seem to be worth.

[00:09:34] Microsoft has invested nearly $14 billion in OpenAI since 2019.

[00:09:39] What role has Amy Hood played in those investments?

[00:09:42] Amy Hood is not by any stretch the engine behind the investment in OpenAI.

[00:09:46] This really was a strategy dreamt up by Nadella and also the company's chief technical officer, Kevin Scott.

[00:09:55] But Amy has been a supporter of it from the start.

[00:09:58] You wrote that some inside Microsoft usually blame Hood for keeping a tight lid on spending.

[00:10:03] So why did she decide that this was one of those areas that was worth such a significant investment, one that she could get behind?

[00:10:10] Well, first of all, it came from the top, right?

[00:10:13] I mean, this is a strategy that Nadella very strongly believes in.

[00:10:17] And so her job as CFO is to help him realize his strategy in a way that makes financial sense for the company.

[00:10:24] The other part is she and the rest of the executive team were genuinely impressed with OpenAI's technology.

[00:10:31] They got a preview of their latest model called GPT-4 in the middle of 2022.

[00:10:37] And so it comes from a place of they believe in it.

[00:10:42] And also they know that everyone else is going to be investing in it.

[00:10:46] And Microsoft was caught a little bit flat-footed by the cloud revolution 10 years ago where Amazon and Google were far out in front.

[00:10:54] And they all were collectively determined to not be left behind this time around.

[00:10:58] And so at the same time that they have been pouring money into building up all this AI technology, they have worked to cut costs in other areas or at least keep a lid on costs.

[00:11:11] Hardware is something that Amy has historically been pretty wary of.

[00:11:15] As Microsoft has tightened down really across the board on spending to make sure that their finances don't go completely off the rails during the OpenAI moment, hardware, from what I hear, has been hit particularly hard or at least been scrutinized much harder than the other divisions.

[00:11:30] That was WSJ reporter Tom Doton.

[00:11:33] And that's it for Tech News Briefing.

[00:11:35] Today's show was produced by Julie Chang.

[00:11:38] I'm your host, Cordelia Janes.

[00:11:40] Jessica Fenton and Michael LaValle wrote our theme music.

[00:11:43] Our supervising producer is Catherine Millsop.

[00:11:47] Our development producer is Aisha Al-Muslim.

[00:11:50] Scott Salloway and Chris Dinsley are the deputy editors.

[00:11:54] And Falana Patterson is the Wall Street Journal's head of news audio.

[00:11:58] We'll be back this afternoon with TNB Tech Minute.

[00:12:01] Thanks for listening.