Inside OpenAI’s IPO Filing, Former SpaceX Engineer on IPO, Apple’s AI Siri Upgrade

欢迎各位收看《信息》旗下的TITV。

Welcome, everyone, to The Information's TITV.

我是阿卡什·帕什拉查。

My name is Akash Pashracha.

今天是6月9日，星期二。

It is Tuesday, June 9.

在正式开始节目之前，明天我会采访几位一直负责报道SpaceX新闻的记者。

Before we get to the show, tomorrow, I'm going to be sitting down with reporters who have been leading the information SpaceX coverage.

他们是我们的埃隆·马斯克记者特奥·韦特、并购记者瓦莉塔·鲍威尔，以及财经副主编科里·温伯格。

Theo Wait, our Elon Musk reporter, Valita Powell, our deals reporter, and Corey Weinberg, our deputy bureau chief of finance.

这场活动仅对订阅用户开放，现在就可以去theinformation.com/events报名。

The event is exclusive to subscribers, and you can RSVP right now at theinformation.com/events.

今天节目中，OpenAI已秘密提交了IPO申请，这预示着它可能迎来历史上最重磅的上市之一。

Today on the show, OpenAI has confidentially filed for its IPO, setting the stage for another of one of the biggest public debuts in history.

我们会与《信息》的OpenAI记者深入聊聊，这对公司意味着什么。

We'll talk with the information's OpenAI reporter for more on what that means for the company.

我们会详细解读苹果全球开发者大会的所有发布内容，包括Siri的AI全面升级。

We'll break down all of the announcements from Apple's Worldwide Developers Conference, including an AI overhaul of Siri.

另外，我们请来了一位前SpaceX工程师，带我们看看本周的IPO，以及他最关注哪些动向。

Plus, we've got a former SpaceX engineer coming on the show to give us a look at this week's IPO and what he's watching for.

节目最后，我们会分享独家报道：大型银行正在探索如何交易计算能力的成本，而期货合约将是这一切的核心。

And we'll close out the show with our exclusive reporting that big banks are exploring ways to trade on the cost of computing power and why futures contracts would be central to all of that.

今天内容丰富，精彩纷呈，咱们马上开始。

It's going to be a great show and a busy one, so let's get right on into it.

OpenAI已经秘密提交了上市申请。

OpenAI has confidentially filed to go public.

这比Anthropic晚了一周，却正好赶在SpaceX即将挂牌上市的同一周。

It is just a week behind Anthropic and comes the same week as SpaceX is about to list its shares.

为我们解读这一切的是我们的OpenAI和谷歌记者王艾琳。

Here to explain to us on all of this is Erin Woo, our OpenAI and Google reporter.

艾琳，欢迎来到节目。

Erin, welcome to the show.

很高兴你能来。

It's great to have you here.

嘿，谢谢邀请我。

Hey, thanks for having me.

嗯，明白了。

Okay.

我们其实早就料到会有这一出。

So we were expecting this to come.

那你觉得OpenAI大概多久能走完IPO的流程？

How quickly can we expect OpenAI to actually move through the IPO process here?

是的。

Yeah.

我觉得不管是Anthropic还是OpenAI，都还得等一阵子。

So I think for both Anthropic and for OpenAI, it's expected to still be a minute.

目前它们只是向证交会提交了保密申请，但离正式上市还有一堆步骤要走。

So right now they've confidentially filed with the SEC, but there's still parts of the process that have to happen before they can actually list.

美国证券交易委员会会对他们的草案提交提出意见，然后他们需要修改，再重新提交。

The SEC is going to give them comments on their draft submission, and then they're going to have to fix that and then they're going to have to resubmit.

所以这可能需要一段时间。

So that could take a while.

我聊过的一个人估计，每一轮反馈大概需要一个月左右。

Like one person that I talked to estimated that it takes like a month approximately per round of common.

不过对于AI公司来说，这个过程可能会更快一些。

And maybe that's faster with AI.

目前我们还不太清楚。

We don't really know yet.

所以这还需要一点时间。

But so it's going to take it's going to take a second.

并不是说他们今天提交了保密申请，明天就能上市。

So it's not like they confidentially filed today and they could list tomorrow.

他们基本上也承认，发布这篇新闻稿的唯一原因，就是预料到会有泄露。

And they basically acknowledged that the only reason they put out this press release was because they expected the leak to happen.

真的吗？

Is that right?

对。

Yes.

如果你熟悉萨姆·阿尔特曼在推特上的风格，这完全就是他典型的推文语气。

It was a very if you're familiar with Sam Altman on Twitter, it was a very Sam Altman on Twitter statement.

其实就是一段话，大意是：嘿。

So essentially, it's just like one paragraph statement saying like, hey.

我们已经秘密提交了申请。

We filed confidentially.

我们料到这事儿会泄露。

We expect this to leak.

所以干脆直接公开了。

We're just gonna announce it.

他们在这份声明里特别想强调的一点是，短期内可能不会上市。

And the other thing that they really tried to emphasize in this statement is the idea that they might not go public kind of anytime soon.

他们其实是在说，如果我们将来决定早点上市，也能有这个灵活性。

Like, they're basically trying to frame this as we want optionality if we decide we want to go public sooner rather than later.

这样就能做到这一点。

This lets us do that.

但与此同时，可能作为一家私营公司，我们还有一些事情想继续做。

But at the same time, like maybe there's things we still want to do as a private company.

所以他们想表达的是：我们正在为上市做准备，但也可能最终不上市。

And so they're trying to position this as we could go public and we're taking the steps, but also we might not.

那么，新闻的第二部分就更有意思了。

So then this is where the second part of the news becomes more interesting.

他们说，员工将在IPO之前获得出售股票的机会。

So they said that they would be looking to sell, well, employees would have the opportunity to sell their ahead of an IPO.

这方面我们了解多少？

What do we know there?

这其实还是为了给自己留更多选择余地。

So basically, this is, again, like trying to give themselves more optionality.

所以 OpenAI 正和 Anthropic 以及其他实验室展开激烈的人才争夺战。

So OpenAI is locked in an intense race for talent with Anthropic and the other labs.

如果 Anthropic 率先上市，那对潜在求职者来说会更有吸引力。

If Anthropic is going to go public first, that makes them more appealing to potential employees.

所以他们这么做是为了给员工提供流动性选择，即便 IPO 的节奏比大家预期的慢，员工也能卖出自己的股份。

So they're trying to do this to give employees the option of having liquidity, being able to sell their shares even if the IPO timing is, like, slower than people want.

有趣的是，这一轮融资的估值将和上一轮私募融资的估值保持一致。

And so interestingly, the valuation of this round is going to be at the same valuation as the last private round.

还是那个8520亿的投后估值。

It's, again, this 852,000,000,000 post money valuation.

所以，基本上他们是在告诉员工，如果你急需用钱，或者想参与旧金山的房地产市场，或者想做其他任何事，这里有个办法能帮你实现。

And so, like, essentially, what they're telling employees is if you need money, if you would like to participate in the San Francisco housing market or, like, whatever they wanna do, like, here's a way to get that.

但如果你想继续持有，想等到IPO时那个更高的估值，那就得等到上市了。

But, like, if you wanna hang on, like, if you wanna hang on for the lot bigger valuation, like, that's gonna wait to the IPO.

好吧。

Well, okay.

说清楚点，我觉得谁都想参与旧金山的房产市场。

To be clear, I think everyone would like to participate in the San Francisco housing market.

我不确定。

I don't know.

不好，因为这正是我看重的地方。

Bad because that's what I appreciate.

明白了。

All right.

好的。

Okay.

我们再回过头来看看关于出售股份的这部分。

Back to this piece here about selling shares.

这是在二级市场上出售，还是公司回购股份？

So would this be selling it on the secondary market or would this be the company buying back shares?

我想把这个弄清楚。

I just want to make that clear.

公司并没有向员工明确说明这件事。

It was not specified to employees.

员工目前对这件事了解得还不多。

Employees do not have a ton of information about this yet.

他们还不清楚自己能卖多少股份。

Like they don't yet know how much they will be able to sell.

他们也不确定这件事具体什么时候会进行。

They don't yet know like what the timing for this is going to be.

如果之后有通知，我也是从一些零散的消息里知道的。

If they've since been told, my signal is in the little thing.

不过，确实还有一些细节还没定下来。

But yeah, so there's still some information that's to be determined.

这其实是OpenAI经常在做的事。

I mean, is something that OpenAI has done a lot.

据我所知，员工们已经卖出了大约一千亿美元的股份。

Like employees have sold around, I think, $10,000,000,000 worth of shares at the lost count.

这并不是他们第一次做这种事，但也不是说从来没人对这个感兴趣。

And so it's not like this is like, Oh, new thing that they're doing for the first time, but it's not like it's something where no one has ever shown any interest in this.

到目前为止，这套流程已经非常成熟了。

So this is a pretty well oiled machine at this point.

没错。

Right.

好的。

Okay.

我总是喜欢提醒大家一下OpenAI当前的财务状况，因为我们经常更新数据，情况一直在变。

And I always like to give folks a bit of a reminder on the current financial state of OpenAI, given that we report things very regularly, the situation changes.

如果OpenAI明天就上市，咱们再回顾一下公司目前的财务状况。

If OpenAI was to go public tomorrow, just remind us of the current financial profile of the company.

是的。

Yeah.

他们最近的年化收入已经突破了300亿美元。

So they've recently passed $30,000,000,000 in annualized revenue.

他们现在显然还在亏钱。

They're obviously still losing money.

运营利润率仍然是负的，我记得上次公布的时候是负122%。

Operating income margin is still negative, I believe, negative 122% last time it was reported.

所以收入是在增长的。

So revenue is growing.

但目前OpenAI的情况是，它已经被Anthropic和它的收入超越了。

The situation that OpenAI is in now though is that they have been eclipsed by Anthropic and revenue.

我们在这档节目里已经聊过很多次了，我就不重复细说了。

Like we've talked about this a ton on the show, so I don't totally need to go into it.

Anthropic通过Cloud Code、Cloud Cowork这类产品，凭借对企业市场的专注，真正超过了OpenAI。

But Anthropic through products like Cloud Code, Cloud Cowork has really been able to surpass OpenAI because of this enterprise focus.

所以现在OpenAI正在努力追赶，特别是在代码生成和企业市场这块。

So right now what OpenAI is trying to do is trying to catch up on codex, catch up on coding, like catch up with enterprise.

因为如果现在这两家公司都上市，就像我的同事们和很多人分析过的，Anthropic对很多公开市场的投资者来说，吸引力要大得多。

Because right now, if the two companies were to both go public, like my colleagues, others have written about this a lot, Anthropic is couldn't be a more attractive bet to a lot of public market investors.

所以这就是OpenAI现在拼命想追上的地方。

So that's what OpenAI is really trying to catch up on.

对。

Right.

但我们现在正处在一个挺有意思的时间点，Codex在很多指标上其实已经比Claude Code更受欢迎了，我们的同事Stephanie Palazzolo就做过不少深入报道，讲大家有多喜欢这个产品。

But of course, we sit in an interesting moment here where Codex is actually now more popular, I think, on many metrics compared to Claude Code, and our colleague Stephanie Palazzolo has done some great reporting on just how people are really liking that product.

所以这更说明我们得经常请你来更新情况，毕竟事情变得太快了。

So that's all the more reason to have you on more often to give us the updates here because things change.

Erin，非常感谢你今天来做客。

Erin, I want to thank you for coming on.

这位是Erin Woo，我们《The Information》负责报道OpenAI和谷歌的记者。

That is Erin Woo, our OpenAI and Google reporter here at The Information.

苹果昨天在开发者大会上发布了全新的AI战略，包括期待已久的Siri升级，以及与谷歌Gemini模型的合作。

Apple unveiled an updated AI strategy at its Worldwide Developers Conference yesterday, featuring a long awaited update to Siri and a partnership with Google's Gemini models.

我想请我们《The Information》负责苹果报道的Aaron Tilley来给大家详细解读一下今天看到的所有内容。

I want to bring on the information's Apple reporter, Aaron Tilley, to break down everything that we saw.

亚伦，欢迎再次做客节目。

Aaron, welcome back to the show.

给我们说说最新情况吧。

Give us the update here.

昨天最大的消息都有哪些？

What was the what were the biggest announcements yesterday?

你印象最深的是什么？

What stood out to you?

是的。

Yeah.

苹果终于推出了全新版的Siri。

So Apple finally unveiled its Siri reboot.

他们把它叫做Siri AI。

They're calling it Siri AI.

这背后是一次重大的战略调整——Siri终于要用更强大的AI模型了，也就是Gemini。

And this the the big sort of strategic shift is so it's finally going to have a more capable underlying AI model, that being Gemini.

Gemini 在幕后扮演着关键角色，支撑着整个模型的运行。

Gemini is playing an important role behind the scenes as far as what's powering the model.

更重要的是，Siri 这次首次变成了一个独立的应用程序。

And then on top of that, Siri for the first time is its own dedicated app.

以前要使用 Siri，要么得喊它的名字，要么得按按钮。

Before interacting with Siri would require either calling her out as the voice part or pressing a button.

这次它就像 Claude 或 ChatGPT 那样，可以打开应用查看历史对话，还能继续之前的讨论。

This time it's a dedicated app similar to a Claude or ChatGPT, where you can sort of go into an app and look at past conversations, look at what else, the ongoing conversations.

从产品策略上看，这确实是苹果的一次重大转变。

And so that strategically as a product is a pretty major shift for the company.

不过昨天我们也看到了一些演示，大概了解了它的能力。

But yeah, yesterday we saw sort of some demos of what it's capable of.

对我来说，最大的感受是，苹果其实把 Siri 的能力说得太低调了。

And a big takeaway from that for me is that Apple really seems to be underselling the capabilities.

两年前，他们可是把功能吹得天花乱坠。

Two years ago, they're really overhyping the capabilities.

我觉得这次他们真的想谨慎得多，不敢轻易承诺这东西能做什么。

I think this time they're really trying to be a lot more measured in terms of what they're promising this thing can do.

我们刚才看到了他们展示的内容。

And we saw what they showed off.

所以你的看法是，他们这波营销策略就是故意低调处理？

So that that's your take on the the marketing tactics here, the underselling here.

如果我们抛开营销，单看实际展示的功能和这东西到底能做什么，你是觉得惊艳还是失望？

If we focus on the capability itself, what they actually showed and what this thing is able to do, were you overwhelmed or underwhelmed in terms of what they what they showed?

说实话，我觉得展示出来的效果挺平淡的，因为……

It's it's pretty underwhelming as far as, like, what the because

我看着的时候也在想，真的吗？

I I was I was watching it, like, really?

这就是所谓的全新Siri？就这？

Like, this this is what it is, the big Siri overall?

我觉得他们现在特别小心，毕竟两年前发布时因为夸大宣传，被起诉了。

I think I think it's they're being very careful because they were sued after the announcement two years ago of false marketing because they were overselling it.

所以我觉得他们在宣传功能时特别谨慎，不敢夸大其词。

And so I think they're being super careful to be measured with what they are claiming and capabilities are.

关于这一点，最有趣的是哪些iPhone能用、哪些不能用苹果智能，以及这一系列新升级的支持情况。

So, one of the most interesting parts about this was which iPhones work and don't work with Apple intelligence and with sort of the new slate of upgrades.

说说看哪些iPhone支持吧，这可是个关键问题——会不会因此带动一波换机潮？

Talk a little bit about which iPhones work, because this is kind of a big question, whether or not this will actually drive an upgrade cycle, right?

确实如此。

Yeah, for sure.

我估计这方面会引发不少争议。

I think there's definitely going to be a lot of groundblading on that.

两年前苹果智能刚发布时，支持的机型范围广得多，这次却要收窄不少。

The Apple intelligence announcement two years ago was able to handle a broad range of phones, and this is going to be a narrower set.

我觉得这背后的原因，是苹果声称设备内存必须足够大，才能运行更复杂的本地模型。

So I think that certainly is is, you know, the part of it is Apple's claiming is the amount of memory on the device capable of running the more sophisticated on device models.

这就是他们的说法。

So that's their justification.

我觉得确实如此，设备内存有限，这确实是个瓶颈。

I think that's, certainly true that there is a limitation of bottleneck in terms of what memory is on the device.

但更根本的是，他们在推动新一代设备的发布，比如九月要推出的折叠屏iPhone。

But for sure, underlying that is a push for new devices, the next generation of devices that they'll announce in September, including their foldable iPhones.

归根结底，他们始终在推动下一轮换机潮。

So it's always about, you know, driving the next upgrade cycle.

我们联合执行主编马丁·佩尔斯昨晚在他的简报通讯里对这篇文章做了非常精彩的点评。

Martin Pierce, our co executive editor, had some great commentary on this piece in his briefing newsletter last night.

我觉得他的结论是，昨天发布的这些升级功能，并不会真正带动换机需求，而这恰恰引出了我们下一个问题：约翰·特尼斯去哪儿了？

I think his conclusion was that the upgrades that were released yesterday, they won't actually drive an upgrade cycle, which at the end of the day, I mean, is what and this is what leads us to our next question, which is where was John Turnis here?

毕竟，苹果还是家iPhone公司。

I mean, Apple is still an iPhone company.

他们的收入全靠iPhone。

They make all their money on the iPhones.

所以约翰·特尼斯就是负责iPhone的那个人。

And so John Turnis is the iPhone guy.

他昨天去哪儿了？

Where was he yesterday?

这是蒂姆·库克最后一次参加WWDC了。

This was Tim Cook's last WWDC.

这是他最后一次参加开发者大会。

It's his last the developer annual developer conference.

所以我觉得，让他安安静静地完成这场属于他的最后一场开发者大会，才是真正的用意。

And so I think just giving him the space to sort of have his last developer conference was the move here.

他确实到场了，也参加了周日晚上那场大型开场活动的周边会议。

He was certainly there and was at sort of the surrounding meetings at the sort of the big sort of unveiling for Sunday night sort of a kickoff event.

从现场出现的频率来看，他哪儿都去了。

So he was all over the place in terms of just being present physically there.

但他确实没在任何发布视频里露面。

But yeah, he wasn't, didn't show up in any of the announcement videos.

我觉得这背后，就是想为蒂姆·库克的最后一次WWDC留点空间。

And I think this is about giving space for Tim Cook's last WWDC.

最后一个问题。

Last question for you.

你之前报道过这件事，但我们现在对苹果与谷歌，还有英伟达的合作有了更多细节。

You had reported this a little while back, but we got a little more color here around Apple's partnership, not just with Google, but also with Nvidia.

昨天在WWDC之外的交流中，我们有没有了解到更多新信息？

Did we learn anything more there yesterday from the conversations outside of the actual WWDC?

你对此有什么看法？

And what was your reaction there?

是的。

Yeah.

在新Siri的主旨演讲中，他们没明说，但在会后活动中，他们承认会依赖谷歌的服务器和英伟达的显卡。

At the keynote for the new Siri, they claimed they didn't say anything, but in the sort of post keynote events, they they acknowledged that they will be relying on Google servers and their Nvidia GPUs.

这和过去的做法大不相同。

And that's a big departure from the past.

2024年发布这些新产品时，他们说所有运算都会在自家服务器和设备上完成。

When they announced in 2024 these new products, they said it would be all running internal servers and on device.

苹果有个叫私有云计算的项目，就是在他们自己的云服务器上运行苹果自研芯片。

Apple has this thing called private cloud compute, which is Apple silicon running in their own server clouds.

这次苹果的芯片确实还跑不动这么庞大的Gemini模型。

And this time Apple, you know, it just isn't capable of running these massive Gemini models.

所以他们只能靠谷歌云和英伟达的显卡来完成这些任务。

So they have to rely on Google Cloud and Nvidia GPUs to do that.

我觉得这对公司来说是个不小的失误。

So that's a big, I think, you know, blunder for the company.

他们之前大力宣传自家芯片有多厉害，运行有多私密、多安全。

The company made a lot of big promises about how great their chips are and how good they, how privately they run.

但现在这些都做不到。

And they can't do that.

目前他们的技术还不够成熟。

They're just not capable enough yet.

所以也只能依赖第三方服务了。

So they're having to rely on third party for that too.

很好。

Great.

艾伦，非常感谢你今天来做客。

Well, Aaron, I want to thank you for coming on.

这位是艾伦·蒂利，我们《信息》杂志的苹果记者。

That is Aaron Tilley, our Apple reporter, here at The Information.

关于WWDC的更多报道，我们接下来请来星座研究公司的首席执行官雷·王，来做一下分析。

For more coverage on WWDC, I want to bring on Ray Wang, the CEO of Constellation Research, for his analysis.

雷，欢迎来到节目。

Ray, welcome to the show.

很高兴你能来。

It's great to have you here.

谢谢你们邀请我。

Hey, thanks a lot for having me.

对于昨天发布的这些内容，你给苹果打多少分？

So what letter grade would you give Apple on this release that we saw yesterday?

我觉得给个B+挺合适的。

I would say it's a solid B.

也就是说，他们2024年承诺的，这次全都实现了。

I mean, it's everything they promised in 2024, it's delivered.

但我也得考虑，要在隐私安全和场景感知、个性化之间找平衡，这确实很难。

But I do have to take into account, I mean, it's hard when you're trying to balance privacy and security with context and personalization.

这两者通常都是互相冲突的。

Those things are typically at odds with each other.

所以这两方面都不能妥协。

And so there's no compromising on those two things.

苹果能做到这一点，已经算是相当不错的B+了。

And for Apple to go and deliver what they were trying to deliver, it's actually a solid B.

B+啊，明白了。

Solid B, okay.

那要怎样才能拿到A呢？

What would have made it an A?

我觉得问题在于，我们总想着AGI、超级智能，觉得它能干所有事，还能预测未来，但往往忽略了那些细枝末节的东西。

Well, I think the challenge is we have this view of AGI, this super intelligence, and being able to do all these things and being predictive, but we also forget that it's the little things on the side.

真正重要的是那些零零碎碎的小操作，得把它们全都串起来。

It's all the little actions that you're trying to bring together.

要是真想拿A，就得有那种让人拍大腿的神功能——比如你突然发现，它居然能猜到你下一步、下下一步、再下一步要干嘛。

And so what would have made an A would be, you know, wham features where you're like, Oh my God, I can't believe it's able to guess three steps ahead of me.

但要实现这个，你就得牺牲大量隐私，或者交出很多个人数据。

But in order to do that, you'd have to give up a lot of privacy or trade a lot of data to be able to go do that.

当然，这些模型现在才刚起步。

And of course, the models are just coming up.

所以借助谷歌的Gemini模型，确实帮了大忙，我觉得他们以后还会不断从中学到东西。

And so leveraging the Google Gemini models is really helping them, and I think they're going to keep learning from that over time.

那现在我想知道，这有没有改变你对苹果整体AI策略的看法？

Now, I wonder, did this change your impression of Apple's AI strategy at large?

我的意思是，这儿的情况跟以前不太一样。

I mean, we have sort of a different beast here.

他们并没有在资本支出上大举投入，去开发新的模型。

They are not going so hard at the, you know, excessive amounts of CapEx, investment in new models.

至少就我们所知，他们的态度相当谨慎。

They're taking a pretty reserved approach to all of this, at least as far as we know.

这跟他们的风格很一致，对吧？

This seemed to be pretty in line with that, right?

你说得对，阿卡什。

You're definitely right, Akash.

实际情况是，他们采取了一种非常务实的做法。

I mean, what's going on here is it's a very pragmatic approach.

想想谷歌、Meta 和亚马逊这些公司，为了建数据中心，要花几百亿甚至上万亿美元，其他公司也在拼命扩建云服务。但苹果却在想：我们真有必要这么做吗？我们已经有12.5亿台设备，分布式AI马上就要来了，这真正受益的是普通用户，而不是那些追求超级智能的公司。

If you think about the hundreds of billions, almost trillions of dollars that's going to be spent on data centers by Google and Meta and Amazon, and of course, all the other companies that are building out the clouds to deliver on that, Apple is like, Why do we need to do I mean, we've got 1,250,000,000 devices, decentralized AI on the way here, and it's really going to benefit the individual and the consumer, not for the super intelligence that everyone else is big.

所以，这是完全不同的打法。

So this is a different play.

苹果可能是这场AI竞赛中最高效的一家，因为他们专注的是你口袋里真正有用的东西。

Apple may be the most efficient player in this AI race because they're fixated on building what's important for you in your pocket.

没错。

Right.

你刚才提到了那12.5亿台设备。

So you talked about the 1,250,000,000 devices.

我昨天看发布会的时候一直在想，AI真的有那么重要吗？

I mean, I guess the thing that I was thinking about yesterday as I watched the keynote was, does it really like, does AI really matter?

AI真能促使人们多买iPhone吗？还是说，它只是个可有可无的卖点？

Is AI really ever going to drive people to buy more iPhones, or really, is it just a checkbox for them?

我离开的时候其实有点疑惑，Apple在AI上做些什么，真的那么关键吗？

I mean, I kind of left thinking, does it really matter what Apple does in the face of AI here?

确实重要，因为我平时根本不用Siri，以前连一天都坚持不下来。

It actually does because I'm not a big Siri user, and I've never turned Siri on for more than a day.

但以后我会重新开启Siri，因为它现在能帮我们完成这么多事情了。

And I will be turning it on going forward because of the ability to do all these different tasks.

视觉智能会变得特别关键。

The visual intelligence is going to be important.

有了这些新的苹果智能功能，不管是写东西、发消息，还是快速切换上下文，我的工作效率至少能提升两倍，甚至三倍。

I'm going get a 2x, maybe a 3x productivity boost, being able to use all these new Apple intelligence features, whether it's writing or composing something, whether it's responding, whether it's being able to move contextually very quickly.

他们举了个例子，说要给朋友办个烧烤派对。

I mean, the example they showed about, Hey, let's do a barbecue for someone.

你家里有什么食材？

Hey, what ingredients do you have?

给我个食谱。

Give me a recipe.

顺便把十五个人都邀请过来参加这个烧烤。

By the way, let's go invite 15 people to that barbecue.

能把这么多任务连起来自动完成，这才是真正的救星。

Being able to thread these multitasks together, I mean, that is the lifesaver.

能自己搞定日常事务的自动化流程，这简直是我梦寐以求的。

And, you know, being able to automate workflows at a personal level, that's a dream come true for me.

好吧，这些功能对你来说是梦想成真，但你觉得人们买iPhone，真会因为Siri好不好用而做决定吗？

But, okay, it's a dream come true for you, but do you think that people will actually make their decision to buy the iPhone or not based on how good Siri is?

这其实正是问题的核心。

I mean, that's the heart of the question here.

我不太觉得Siri好或不好，会真正让那些想加入苹果生态的人打退堂鼓。

And I'm not so convinced that Siri being great or bad is really going to dissuade people who want to participate in the Apple ecosystem.

不是吗？

No?

你说得对，但Siri好也能防止用户离开苹果，去用别的厂商AI功能更强的产品。

You're right, but it also keeps them from not leaving Apple to go to someone else that might have a better feature in AI.

这方面就这一点影响。

That would be the only thing on that end.

所以它确实让用户更难离开。

So it definitely keeps it more sticky.

它肯定不会让安卓用户转投iPhone。

It's definitely not going to switch people from, you know, an Android phone to an iPhone.

但关键在于，它能帮助苹果推出更多服务，推动整体市场规模增长，靠的是开发者能打造的新功能。

But the main point is it's going to actually allow them to build more services and drive TAM up based on those services that developers are able to create.

我们知道，这是蒂姆·库克最后一次参加WWDC，约翰·特尼斯将在秋天接棒。

Now, we know that this was Tim Cook's last WWDC, and John Turnis will take the reins in the fall.

你觉得在约翰·特尼斯接手后，苹果对AI的策略会有变化吗？

Do you have any view on whether or not AI or the approach to AI will change at all under John Turnis?

我觉得策略不会变，但心态会不一样。

I think the approach isn't going to change, but I think the mindset will change.

蒂姆·库克的任务就是把现有的东西整合好，想办法推出下一个重大创新，同时扩大总可服务市场、提升收入、增加利润率。

Tim Cook's job was really to take the existing pieces, try to figure out how to come up with the next big innovation, and really grow TAM, grow revenues, and grow Warchast.

而蒂姆确实做到了。

And that's what Tim did.

他是个效率专家，供应链高手，能灵活调整定价、推动涨价，这些举措彻底改变了苹果的现金流状况。

Mean, he was the efficiency expert, he was the supply chain expert, he was able to actually drive price increases, adjust pricing as needed, and that really changed the game for Apple, being able to have the cash flow.

现在，我们需要一个更关注产品、创新和工程的人来做这些决策。

Now we need to go back to someone that's focused on product and innovation and engineering and make those decisions.

苹果最大的挑战不是该做什么，而是不该做什么。

Apple's toughest challenge isn't what to build, it's what not to build.

他们其实能做很多东西，从他们搞电动车那件事就能看出来。

They could build a lot of things, and you can see that from their exercise in building out the EV.

他们很清楚，这是一场必须冲到底的竞赛。

I mean, they realized this was a race to the end.

中国电动车的价格都降到七千美元一台了。

I mean, Chinese EVs are at $7,000 a hop.

苹果做的是高利润、高销量的生意，电动车这条路走不通。

Apple's in the high margin, high volume game, and this wasn't going to take them there.

所以希望有了约翰，他们能真正找到iPhone之后的下一个重磅产品。

And so hopefully with John, there's going to be that ability to actually find that next big thing after the iPhone.

那现在有哪些可能的方向呢？

What are the options here?

我们听说，翻盖手机好像有点苗头，从一些设计布局上能看出端倪。

We've heard I mean, flip phone, I guess, was sort of teased in some way with the layouts.

我知道很多人其实也说不准。

I know people were really I don't know.

当时好像很多人都在分析，哪些软件能用在翻盖手机上。

It seemed like there was a lot of analysis on what software could work on flip phones.

我不觉得这算得上是最大的赌注，毕竟这事儿是在蒂姆·库克任内一直在做的。

I wouldn't really call that the biggest bet ever, certainly because it's been worked on under Tim Cook's tenure.

但说到重大押注，你觉得特恩斯在未来几年可能有哪些大方向？

But when you talk about big bets, what are the menu of big bets that you think Ternus could take in the next few years?

技术会不会发展得足够快，让视觉探测设备达到像Meta眼镜那样的水平？

Is the technology going to advance quickly enough that a vision probe gets down to where meta glasses are?

会不会出现另一个健康类设备，能以前所未有的方式实现精准健康管理？

Is it going to be another device that sits at health that's going to give you the ability to do quantified health in a way that we hadn't seen before?

会不会有某种植入式设备或装置，来增强我们的能力？

Is there maybe an implant or device that's used to augment our ability to do things?

会不会是个能自主行动的个人机器人？

Is it a personal robot that pumps up?

可能性太多了，而苹果的风格，一向是先开创市场，再打造独一无二的用户体验。

There's so many choices that are out there, and of course, it's Apple's way of launching a market, launching a user experience that makes it unique.

没错。

Right.

我想问问你对苹果依赖谷歌和英伟达的看法。

I want to ask you about Apple's reliance on Google and on Nvidia.

昨天也提到了这个话题。

That was a theme that also came out yesterday.

你觉得他们不得不依赖外部合作伙伴，你怎么看？

What's your reaction to them having to rely on outside partners at all?

我觉得有必要把两件事区分开来。

You know, I think it was an important step to distinguish two things.

一方面，他们用的是最先进的模型，不管是 Gemini，未来也可能换成 Anthropic，或者别的公司。

One, they're using the most advanced models, whether it's Gemini, you could drop it out for Anthropic in the future, you you could swap it out for someone else.

但对企业的需求来说，谷歌是个不错的选择。

But Google for the enterprise is a good choice for them.

另一方面，是他们如何处理隐私和信任的问题，对吧？

And the second piece is that how they're handling privacy and trust, right?

他们有自己的底层系统，需要调用外部的通用模型或其他高级模型时，会确保上下文和隐私安全，再把结果和洞察带回内部。

They've got their own layer, and then when they need to go out to world models or other types of advanced models, they'll go out, they'll protect the context, they'll protect the privacy, and then bring back the information and insights.

我觉得这才是关键所在。

I think that was the important piece.

他们并没有放弃对隐私和信任的坚持。

It's they're not giving up on privacy and trust.

这也解释了为什么他们花了这么长时间。

And it's also why it took them this long.

很好。

Great.

雷，非常感谢你今天来做客。

Well, Ray, I want to thank you for coming on.

这位是恒星研究公司的首席执行官雷·王，正在TITV节目中。

That is Ray Wang, CEO of Constellation Research, here on TITV.

我们昨天在节目中聊过SpaceX的那群元老，就是那些离开后自己创业的前员工。

We spoke yesterday on the show about the SpaceX mafia, or the former SpaceX employees who have gone on to start their own companies.

他们很多人在公司期间获得的股份，现在都价值不菲。

Many of them stand to gain big from the shares they earned during their time at the company.

其中一位前员工是斯科特·马丁。

One of those former employees is Scott Martin.

他是Revel公司的创始人兼首席执行官，这家公司开发工业软件，最近估值达到12.5亿美元。我想请斯科特谈谈他在公司的经历。

He is the founder and CEO of Revel, a company that makes industrial software, which was last valued at $1,250,000,000 I want to bring on Scott to reflect on his time at the company.

斯科特，欢迎来到节目。

Scott, welcome to the show.

很高兴你能来。

It's great to have you here.

是的。

Yeah.

能来我很高兴。

Great to be here.

好的。

Okay.

真是忙碌的一周。

Big week.

你感觉怎么样？

How are feeling?

我感觉棒极了。

I'm feeling awesome.

对啊。

Yeah.

这可是个真正的客户，你知道的，我这边……

This is, you know, I'm yeah, a real client for

群聊那边都热火朝天了吧？

chats are the group chats are are firing up over there?

是的。

Yes.

我们内部的聊天群已经炸了。

We've got internal chat here.

Rebel公司招了不少前SpaceX的工程师。

Rebel, went to former SpaceX engineers.

大家特别兴奋。

Super excited.

我理解得对吗？你们公司大部分——其实很多员工，都是前SpaceX的工程师？

And do I have this right that most of your company I mean, a lot of your company, are former SpaceX engineers, right?

是的，没错。

Yeah, that's correct.

我们大概70%都是前SpaceX的软件工程师。

We're, yeah, about 70% are former SpaceX software engineers.

真厉害。

Wow.

你能说说吗？SpaceX的文化里，有哪些东西是你在自己公司里特别想借鉴的？不管是文化、流程，还是习惯和做法？

And tell me, culture at SpaceX, what were the things from SpaceX that you have really tried to emulate in your own company insofar as culture or even routines, habits, patterns?

对。

Yeah.

我觉得，这其实是高自主性与高责任感的结合。

I mean, I think, yeah, I think it's a combination of like high agency but high accountability.

我们确实继承了SpaceX那种注重执行力的文化。

So we've got we definitely have kind of the SpaceX strong execution culture here.

还有就是以使命为导向的思维方式。

But, also, I think, like, mission the mission oriented approach.

我们的使命是帮助那些正在研发极其复杂项目的企业提速，让他们能走得更快。

Like, you know, we're our mission here is to accelerate all the companies out there, building extraordinary extraordinarily difficult things and really enabling them to move much faster.

你可能不太清楚Revel具体是做什么的，我们是一个用于控制和测试硬件系统的软件平台。

I don't know if know about Revel, what exactly we do, but it's a software platform for controlling and testing hardware systems.

比如火箭发动机测试场地、核反应堆，甚至你家里的自来水系统，背后也是一连串由软件控制的泵和阀门。

So think like rocket engine test sites, nuclear reactors, even how the water actually gets to your house is a series of pumps and valves that have a lot of software involved.

很多公司正在研发像高超音速飞机这样前沿的系统，但他们用的软件却还是上世纪80年代的老古董。

And and you kind of have these, like, these companies that are building extraordinarily complex things and cutting edge systems like hypersonic jets, and they're using software that dates back to like the 1980s.

那可是比Windows 98还早的软件了。

So this is like pre Windows '98 era software.

所以Rebel就是来解决这个问题的。

And so Rebel is here to fix that.

没错。

Right.

对。

Right.

那从SpaceX那里，有哪些东西你们决定不带过来，或者不想在公司里继续效仿的？

What about things from SpaceX that you are deciding to leave at SpaceX or things that you don't want to continue emulating at your company?

我在SpaceX度过了非常美好的时光。

You know, I had a great, great time at SpaceX.

我们确实取得了了不起的成就。

I think it was you know, we achieved incredible things.

我当初加入，就是想参与这个了不起的使命——让火箭可重复使用，最终实现登陆火星。

You know, I originally joined because I just wanted to work on this incredible mission, you know, making rockets reusable and ultimately getting to to Mars.

是的，没错。

And, you know, I yeah.

我觉得SpaceX是个很棒的工作地方。

I don't I think SpaceX is a great place to work.

我曾经在那儿……

I had

不，不是的。

to No.

听我说，我并不是在寻找什么。

And look, I'm not I'm not looking for anything here.

我只是有点好奇，你看那些从SpaceX出去、自己创业的员工，数量这么多。

I just you know, I I sort of wonder You have all these SpaceX employees that have now branched out and started their own companies.

我们已经听过太多关于他们超强的工作态度，以及对宏大目标的坚定信念。

And I know we've heard so much about the tremendous work ethic, the ability to believe in a moonshot mission.

不过我觉得，可能也有一些细节是我们不希望复制的，比如那种不惜烧钱好多年的做法。

You know, I also think that there are probably some nuances in the way that, you know, maybe we don't want to create a company that won't burn money for X amount of years.

也许我们更希望把财务目标定得更务实一些。

Maybe we want to go a little narrower in terms of our financial profile.

我不确定。

I don't know.

是的。

Yeah.

我的意思是，确实如此。

I mean, I think yeah.

我觉得吧，SpaceX这地方不是人人都适合的。

I think it's a you know, I think it's not for everyone, SpaceX.

那里的人工作都特别拼命。

Like, you know, people work extremely hard there.

时间安排非常紧张。

The timelines are pretty tough.

但我觉得，这正是取得成功所需要的。

But I think that is kind of what it takes to be successful.

没错。

Right.

如果你回望你在那里的时光，你待了那么久，不妨花一点时间想想。

If you look back on your time there, and you were there for a long time, and you could take a second here.

我很好奇，在 SpaceX 那么多年里，最让你印象深刻的回忆是什么？是哪一天让你至今难忘？

I wonder what the most defining memory you have there, what day you remember at SpaceX among all the years you spent there?

有没有哪一天、哪一个瞬间，特别清晰地留在你记忆里？

What day or what memory really sticks out for you?

对，确实如此。

Yeah.

我刚开始是做发射操作的，当时我写了猎鹰九号的推进剂加注流程，就是把燃料从发射台输送到火箭上的那个程序。

Mean, I think so I started working on in launch operations, I actually wrote the propellant load sequence for Falcon nine, which is what, you know, takes propellant from the the launch pad and puts it on the rocket.

我觉得我参与的第一场发射，我本人就在控制室里。

And, you know, I think that first launch that I was a part of, I was actually in the control room.

我记得那天早上走进去，环顾四周，整个团队年轻得惊人。

And I remember I walked in the morning, and I looked around, and it was like it was such a young squat.

当时我印象里，超过三十岁的可能只有发射指挥官一个人。

It was like I think the only person over 30 was the launch director.