投资者的关注点转向利率和人工智能

欢迎收听《市场观点》。

Welcome to thoughts on the market.

我是摩根士丹利全球跨资产策略研究主管唐塞拉。

I'm Serena Tang, global head of cross asset strategy research at Morgan Stanley.

今天，我将和大家分享我最近在欧洲和亚洲与投资者会面的总结，重点聊聊人工智能和美联储的最新讨论。

And today, I'm bringing you a debrief from my investor meetings across Europe and Asia and the key debates around AI and the Fed.

现在是纽约时间7月2日，星期四上午10点。

It's Thursday, July 2 at 10AM in New York.

过去两周，我一直在欧洲和亚洲各地走访，与投资者交流摩根士丹利的最新观点。

The last two weeks, I have been traveling in Europe and Asia to meet with investors to discuss Morgan Stanley's latest views.

我走到的每一个会场，几乎都围绕着两个话题展开。

Two themes dominated nearly every room I walked into.

第一个是美联储以及美国的货币政策走向。

The first is the Federal Reserve and monetary policy path in The US.

许多投资者把凯文·沃什主席在六月FOMC会议上的首次表态，解读为明确的鹰派立场。

Many investors had interpreted Chair Kevin Warsh's June FOMC meeting, his first at the helm, as unambiguously hawkish.

我所接触的投资者指出，美联储的经济预测摘要（简称SEP），也就是对GDP增长、通胀和联邦基金利率等宏观指标的预测，显示2026年可能还会加息，并推迟降息，意味着政策将保持更紧的状态。

What market investors at my meetings pointed out is that Fed's Summary of Economic Projections, commonly shortened to SEP, which details policymakers' forecasts for macro metrics like GDP growth, inflation and the federal funds rate at a hike in 2026 and pushed out rate cuts, implying more restrictive policy.

不过，摩根士丹利的经济学家认为，六月FOMC会议SEP中暗示的加息，需要谨慎解读。

Now Morgan Stanley's economists think that hikes implied by SEP at the June FOMC meeting should be interpreted with caution.

这些预测基于近期通胀居高不下的前提，可能没充分考虑到经济全面重启后通胀自然回落的趋势。

The projections appeared conditioned on elevated near term inflation and may not capture the disinflation from a straight reopening.

我们预计核心通胀会走低，因为旅游相关通胀回落加上关税带来的价格回调共同作用，所以我们认为美联储到2026年都会维持利率不变。

We actually anticipate a lower path for core inflation, given a combination of a reversal in travel related inflation and tariff payback, which lead to our call that the Fed remains on hold through 2026.

另一个在各地投资者会议中反复出现的主题，不出意外，就是人工智能。

The second recurring theme in meetings with investors across regions is unsurprisingly AI.

虽然每次会议投资者都坚定看好AI资本开支的持续周期，但也有人感到不安，因为AI现在不仅在宏观层面成了通胀因素，在微观层面也成了融资问题。

While in every single meetings, investors believe firmly in the cycle of story of ongoing AI CapEx cycle, there was some unease, especially since AI is now also becoming an inflation story on the macro side and a funding story on the micro side.

‘芯片通胀’成了新词，市场还在争论它是否会成为拖累AI资本开支周期的因素之一。

Chipflation is a new word in town with markets still debating whether it can be one of the things that derail the AI CapEx cycle.

在我们的经济学家和行业分析师看来，情况要复杂得多。

In our economist and sector analyst views, it's more nuanced.

虽然过去一年内存价格涨了六倍，但我们认为芯片涨价更可能带来AI基础设施的重新定价和配给，而不是打断整个周期。

While memory prices is up six fold over the past year, we think Chipflation is more likely to reprice and ration AI infrastructure than derail the cycle.

AI的需求正在三个层面同时扩张。

AI demand is scaling across three layers at once.

每块芯片需要更多内存，每个系统需要更多芯片，每个集群需要更多系统，而超大规模云服务商始终优先获得供应。

More memory per chip, more chips per system, and more systems per cluster, while hyperscalers remain first in the allocation queue.

现在，关键风险在于资本支出的效率。

Now, the key risk is CapEx efficiency.

内存在AI系统总成本中的占比正在上升。

Memory is becoming a larger share of the AI system cost.

但我们认为这个周期依然稳固。

But the cycle we think remains intact.

至于AI的资金需求，投资者一直在讨论它还能加速到什么程度。

As for AI funding needs, the debate with investors has been how much more can it accelerate.

值得注意的是，今年至今企业发行的债券中，大部分都是用于数据中心建设的融资。

It's worth noting that the majority of corporate bond issuance quarter to date has been related to funding construction of data centers.

大型云服务商一直在通过非美元债券来扩大投资者基础。

Hyperscalers have been broadening their investor base through non dollar issuances.

今年五月，它们合计发行了约250亿美元的欧元、瑞士法郎和日元债务。

They have collectively issued around 25,000,000,000 of debt in other currencies like euro, Swiss franc, and the JPY in May.

我们的信贷策略团队预测，2026年全球与AI相关的债券发行量将再增加近6000亿美元。

Our credit strategy colleagues forecast nearly another 600,000,000,000 of AI related global issuance in 2026.

这意味着对美国来说。

Meaning for U.

对美国而言。

S.

仅美国投资级企业债一项，我们预计今年净发行量将达到1万亿美元，这也是我们看淡这一资产类别今年表现的原因。

IG corporate bonds alone, we expect 1,000,000,000,000 of net issuance, a reason for our view that the asset class can underperform this year.

结合我们股票团队的估算，到2027年，大型云服务商的现金资本支出将突破1万亿美元，我们预计发债速度还会加快。

With our equity colleagues estimating hyperscalers cash CapEx to surpass 1,000,000,000,000 in 2027, we expect issuance to accelerate.

总的来说，全球投资者都在应对美联储政策和AI资本支出带来的相同不确定性，这些议题在未来仍将是核心讨论焦点。

Bringing it all together, investors globally are all grappling with the same uncertainties around the Fed and AI CapEx, which will likely continue to be key debates to come.

但摩根士丹利的基准预测认为，通胀下行将促使美联储维持利率不变，同时AI资本开支周期依然强劲，因此我们建议投资者继续对风险资产保持积极态度。

But Morgan Stanley's base case view of lower inflation driving the Fed to stay on hold and a strong AI CapEx cycle that remains intact means we recommend investors should still stay constructive on risk assets.

感谢收听。

Thanks for listening.

欢迎留下评价，告诉我们你的看法。

Let us know what you think by leaving a review.

如果你喜欢这个播客，今天不妨和朋友或同事聊聊你对市场的看法。

And if you enjoyed the podcast, please share thoughts on the market with a friend or colleague today.

本内容不构成任何投资要约或劝诱，也不构成税务或法律建议。

It is not an offer or solicitation nor is it tax or legal advice.

我们未考虑您的财务状况和投资目标，可能并不适合您。

It does not consider your financial circumstances and objectives and may not be suitable for you.