跳到正文
Semafor· Shelly Banjo·· 3 小時前AI 評分49

AI 投資熱潮正撞上資金緊縮

The AI bet is running straight into a capital crunch

AI 導讀

DBS Group 行政總裁 Tan Su Shan 警告,AI 建設熱潮將面臨資金短缺,放貸方已開始要求實際回報。她預計,AI 供應鏈企業未來兩年或仍可維持良好狀況,其後 AI 用戶將要求投資回報,但回報未必能實現。投資者正更審慎地檢視 ROI 和資本開支,Tan 認為市場要求與風險相稱的回報並非壞事。

正文

The Scene

A capital crunch is coming to the AI buildout that’s already plagued by shortages of chips, power and materials.

“Soon, there will be shortages of capital” and lenders are starting to demand real returns, Tan Su Shan, who runs Southeast Asia’s largest bank said at a Semafor World Economy Global Leaders Dinner in Bangkok.

“When the party will end, I don’t know,” the CEO of DBS Group said at the event, held on the sidelines of the World Bank and IMF annual meetings. But “there will come a point where people need to question the returns.”

Her bet: If you’re anywhere on the AI supply chain, you might be okay for, say, the next two years, but probably not longer than that.

Global investors are already showing “jittery nerves,” Tan said, following a chip selloff across global markets sparked by the Broadcom CEO Hock Tan’s comparison of AI investors to casino gamblers. A lot of investors have “no clue what they are investing in,” he said at DBS’s tech conference last week.

But the market reaction isn’t necessarily bad, Tan argued. Unlike during the global financial crisis in 2008, she said, investors can see the weakness coming and that’s starting to play out with more discerning buyers demanding risk-adjusted returns.

“They’re bringing up this ROI question, the capex question. The market is speaking, and I don’t think it’s a bad thing,” she said.

Q&A

Shelly Banjo: The CEOs of Broadcom and Intel sounded the financial alarm on AI, likening investors to casino gamblers. Do you share that concern?

Tan Su Shan: There’s shortages in memory. There’s shortages in chips. There’s shortages in power. And soon there will be shortages in capital. So if you’re anywhere around that supply chain, the next two years you’re going to be okay. But what will then happen is that the users of AI will start demanding returns. Now those returns, the ROI, may or may not materialize.

Hock [[Tan]] said, “Well, some people are just investing like it’s a casino. They don’t really know what the returns are, but they’re just punting away.” I can’t remember his exact words, but, sort of, it’s a casino. People don’t know what they’re doing. It’s like mom gave you some money to go play video games, and you’re playing video games and wasting the money, and I think Elon Musk, latched on to that, he said, “No crying in the casino.”

It really spooked the markets. Why do you think it had such an outsize impact?

There was quite a lot of reaction. It does speak to the jittery nerves in the market. But the fact that we’re all talking about it is actually not a bad thing. Unlike previous crises where they just came upon us quite suddenly in one fell swoop, people are now bringing up this ROI question, the capex question. The bond market has spoken, right? I’m demanding more returns for the kind of risk that we’re taking, both on the AI front and the US fiscal deficit. So I think the market’s speaking and it’s not a bad thing.

It’s not like there isn’t a lot of money in the world, though. So where do these capital fears come from?

The US capital markets have convinced the whole world to invest in their growth companies. No matter how much money you’re losing, it’s okay. In the bond market, no matter how much you’re spending, it’s okay. The incredible genius of the US is its ability to hoover up investments from everyone — from sovereign wealth funds, to private individuals, to retail, down to the youngest investor. The kids are now in Robinhood buying SpaceX.

But with this massive hoovering of dollars into this concentrated bucket, there will come a point where people need to question the returns. When the party will end, I don’t know.

The private credit market is also adjusting, which is not a bad thing. So I’m hopeful that we could see an equilibrium.

How are you thinking about the projects that are locking in this long-term debt at higher rates, and how do you think about that risk for your bank and the customers that you’re serving?

The market is adjusting and a lot of the hyperscalers are tapping the market before their credit ratings get downgraded. But they’re moving from being massive cash generators to becoming massive cash guzzlers, and that in itself will be a massive pivot.

You’re sitting right in the center of all these geopolitical fault lines, with business in Taiwan, mainland China, and across Asia. How do you strike a balance?

This is a great time for middle powers. There’s two big powers, and then there’s everybody else. And all these middle powers, everyone has this ability to stay open, stay world-based, and continue to transact with each other. And frankly, the better connected we are, the safer it will be for everyone. Why? Because AI unleashed with no guardrails and no sharing of best practices and no sharing of cyber know-how means no one’s going to win. This is something that the world needs to take seriously.

You’re channeling your inner Mark Carney here.

That speech was outstanding. We all need to work better together. This is not a time to start being selfish.

DBS was early on using AI with your wealth advisers. Are you at a point where your massive army of wealth advisers are now taking orders from the robots? And if so, can’t your customers just build their own?

They probably are, and that’s fine. And at some point in time, it may be our agents talking to customers’ agents. But the key is you want to be there for when their agent’s looking for stuff, you want to be so embedded in their entire journey. Because it should be trusted more than something out on the internet, grounded in policies and governance. And there should be guardrails around what we can and cannot do. Where someone is looking and evaluating your agent. Someone who’s making sure that agents are not accessing data they shouldn’t be accessing, that they’re not doing stuff they shouldn’t be doing, and that you’re evaluating, ‘are they doing what we want them to do?’ Because if they’re not, sayonara, kill the agent.

來源:Semafor · semafor.com