원문 보기
Tom Lee and Dan Ives make big predictions about the stock market. Tom is talking about the S&P 500 reaching 16,000 points. Dan believes that a multi-year " bull" technology market lies ahead, where demand for AI chips exceeds supply. But how do they go from building data centers to making such large-scale predictions ?
Fundstrat market strategist Tom Lee spoke with Dan Ives , a longtime tech bull , at the Future Proof event in Huntington Beach, California. These were their predictions, and Tom simply explained why he thought the market could continue to grow . He noted that company profits rose by about 25 % that year, while the market rose by only 10%.
Prices rose, but profits rose even faster. You pay more for stocks, but you get more return on every dollar spent. So a growing market may be getting cheaper, which is hard to wrap your head around. There are two paths that can push stock prices up. Companies may earn more, or investors may pay more for each dollar of their profits.
Dan believes that the demand for AI will bring more business. Tom also suggests that investors may value these earnings higher. Both scenarios are possible, but business development takes time, while investor sentiment changes quickly. This is what Tom's forecast for the Fed was based on. He believed that a rate hike would reassure investors by convincing them of the Fed's actions, which would reduce expectations for further hikes.
If the Fed takes this step, he predicted a very high probability of a powerful rally starting the very next day. His forecast hinged on investor relief, but the Fed's own forecasts cast doubt on that. On September 16, 2026, the Fed raised the target range by a quarter of a percentage point to 3.75–4%, and most officials expect further increases through the end of the year.
The Fed acted, but that didn't mean everything was nearly over. Therefore, it was necessary for investors to see this first increase and worry less about the next one. Tom himself puts the Fed almost at the top of the list of risks. When asked about the most likely bearish scenario, he named two things that kill a bull market: a real bubble and Fed intervention.
According to his estimate, in 80% of cases the responsibility lies with the Fed. So his optimism depends partly on how far the Fed will go. He believed that one hike would bring relief, but further tightening remains a risk to his bullish forecast. Short announcement: thank you to WhisperFlow for sponsoring this video.
Since the birth of my child, the ability to speak instead of type has been a real lifesaver for me. This is a voice-to- text tool, and now I'll show you what I mean. Look. Hi Anya, can you send the draft versions on Thursday? Actually, on Friday. Please check three things: introduction, subtitles, and sound.
So, as you saw, I changed the day on the fly, and the program fixed everything itself. The most useful thing here is that I can literally think out loud without having to make every correction manually. I started using it when I was holding a baby, but now I use it for everything: letters, messages, and longer texts with context for Codex and Claude.
This also works inside the apps I already use. Help me plan tomorrow's shoot. I need to record the main video, film a sponsor demo, and leave time for a sound check before sending everything to the editor. And I can use WhisperFlow on my phone too. I will be home at six. Actually, it's half past six.
Please leave my dinner in the fridge. WhisperFlow is free, so there's no reason not to try it. To get a month of WhisperFlow Pro with unlimited words, use the link in the description or enter DC in the plans and payment section on your computer . When asked where you learned about this, select YouTube and enter David Carbon.
Tom's long-term confidence is based on the companies themselves. He noted that executives adapted to shocks and cut costs, describing the state of top managers as a permanent crisis mode. He believed they had proven their ability to cope when conditions became difficult. Cutting costs can give a company more profit from the same sales, but there is another path to growth.
Sell more than what customers need. This takes us from the question of investor reaction to the Fed to what these companies can actually sell. Dan believes they can sell in an AI shortage that will last for years. Dan estimates demand for AI chips to be 13 to 1 relative to supply. According to him , for every chip produced, there are 13 orders.
He says he sees this pressure firsthand in Taiwan and Korea. At this rate, he doesn't expect supply and demand to balance until early 2029, which sounds absolutely crazy, right? Note that this is Dan's estimate, not the number of orders that we have independently verified . His argument is that the suppliers' customers may have to wait years.
If you find this content useful, click " subscribe". As you've already seen, this helps me invest more in improving these videos. But once all this AI infrastructure equipment arrives, the customer will need to get it up and running. Buying chips may allow a business to launch an AI service, but people still need a reason to use it.
If the service works and customers continue to pay, this business has a reason to expand. Here's how these purchases can lead to new orders. Dan believes the costs will go far beyond the chips themselves. For every dollar of capital expenditure, that is, money spent on building and equipping a business, he estimates a multiplier of five to six dollars for the rest of the technology sector.
He points to software, infrastructure, and cybersecurity. Equipment needs software to operate, and businesses need to protect their systems. So his argument includes the companies that provide these services as well as chip manufacturers. The five to six dollar figure remains his estimate of the costs in the technology industry , not the money that each supplier is guaranteed to receive.
Second, Dan also expects that 1,200 to 1,500 data centers will be built in America over the next five to six years . He suggests that some projects may be rejected, but believes that the development is just beginning. This forecast requires two things: customers who need the capacity and an industry capable of building it.
The shortage helps explain demand, but it can also complicate the construction process. Tom described annual capital spending at around $800 billion, moving towards the $1.1 trillion mark. He discussed the industry's efforts to establish its own regulations. We will return to why this is important, because a client's desire to obtain more computing power does not automatically mean building a new data center.
But if the opportunities are so great, why are investors paying less for every dollar of profit? This is what helps us understand the price-to-earnings ratio (P/E). It compares the market price to the profit behind it , and people call it a multiplier. At a 22x earnings multiple, the price is $22 for every $1 in annual earnings.
This gives us a way to compare price and earnings, instead of just looking at whether the stock price has increased. Imagine we started with the same multiplier of 22 and applied Tom's growth rates. Prices increase by 10%, but profits by 25%, so the multiplier drops to about 19.4 times profits. Companies are making more money, but prices are not keeping up , and that leaves us with a useful question.
Why aren't investors willing to pay the same amount for every dollar of these higher returns? Memory chip makers are showing us why investors may be wary. A shortage can push prices up and make profits huge . But if investors expect this deficit to end, they can also expect earnings to fall . Today's gains look good, but they may not last long.
So a low multiple doesn't necessarily mean investors missed an opportunity. They may understand this perfectly well , but believe that high profits are temporary. Tom believes that Nvidia may be misjudged just like Apple once was. He says investors didn't value Apple properly because of the App Store.
According to him , Apple's multiple almost doubled as investors changed their view of the business. A company doesn't have to double its profits for investors to double the amount they're willing to pay for each dollar of revenue . In this comparison, the change comes from greater confidence that these gains will last for a long time.
These are the kinds of changes that Tom believes could happen with Nvidia. Tom compared Nvidia with a multiple of around 16 and Costco with a multiple of around 50. He joked that if Nvidia charged membership fees, its multiplier would go up to 50. I like that because it's immediately clear what he means .
He believes that investors value more profits that they believe will come again and again. This doesn't prove that Nvidia deserves the Costco multiplier, but it does make the issue much clearer. How long do investors think good returns will last ? Dana approaches this question through earnings forecasts.
He believes that investors are underestimating earnings by about 25% over the next few years. He expects companies to exceed these expectations quarter after quarter. If this happens, the price paid by investors will look lower relative to the business's actual profits. His argument requires companies to make higher-than- expected profits, not just everyone becoming more optimistic.
Better results would reinforce this enthusiasm. Tom also noted that fund managers had little or no stock because they considered it a " bubble." In his opinion, this leaves more room for institutional investors to enter the sector. This doesn't mean they will change their minds or be forced to buy, but it brings us back to two big variables.
Companies can earn more, and investors can pay more for those profits . One does not guarantee the other. The business can continue to grow as long as investors remain cautious about the duration of that growth. Tom's prediction of 16,000 came about during a discussion of well-known companies. He believed that some of their benefits might last longer than investors expected.
Companies that benefit from new technology don't necessarily have to be new. Some already have clients who may need more than what they can offer. When asked about Costco and Dell, Tom said investors wrote off companies like Cisco years ago. Tom gave an example of their existing customer base. A new competitor cannot simply recreate this relationship by coming up with a promising product.
So, AI can give sustainable businesses more work to do, instead of automatically handing off their customers to someone else. This is an advantage that Tom believes investors should take seriously. Dan expressed concern that AI could destroy software companies. The software stock sell-off has been a big topic this year, and Dan called the idea behind it a fairy tale.
He pointed to Anthropic's entry into the cybersecurity field. One example is their security tool Claude, which can scan software code for vulnerabilities that hackers can exploit and then suggest fixes. Dan recalled that around April, when Anthropic released a security product, people decided that the existing companies in this field were finished.
Dan saw this as a story that investors had invented for themselves, not the end of companies that protect business systems. More computing power can create more work around it, so he includes software in investment opportunities. He argues that this business can benefit from AI, rather than believing that it will make them all redundant.
But the queue for chips doesn't tell us whether each funded project will bring in a profit. The host of this conversation between Dan and Tom pointed out companies borrowing money and selling stocks, and Tom pointed out that for a debt bubble to occur, money must be too cheap. He said that financing cost rates are rising, putting more pressure on weaker projects.
This may scare away low-quality projects, but it also leaves good companies with financing costs to cover. Imagine a business purchasing equipment to provide AI services. The equipment manufacturer gets the sale before the buyer even knows if enough people will use the service. If customers continue to pay and the business covers its costs, there is a reason to expand.
If not, ordering equipment is harder to justify. That's why a full order book today doesn't tell us the whole story. Tom described another risk: when everyone decides that there is only one AI model worth using. In such a scenario, he said, capital expenditures could drop to zero . This is his hyperbolic hypothesis.
The question is whether customers will continue to want to choose from a range of systems , or will they focus their demand on one. This matters to an industry that spends huge amounts of money to service them. Dan expects hundreds of models to eventually become commodities, meaning greater interchangeability and price competition.
It places greater emphasis on companies controlling their own data rather than allowing large model developers to access it. This is part of what he means by sovereign AI in this discussion. Different businesses may want different terms for their data, but if models become easier to replace, it will be harder for the vendor to charge a higher price.
Cheaper AI models can help the customer while putting pressure on suppliers' profits. And before the industry can meet all this demand, it needs permission to build facilities. Dan cited the politics surrounding data centers, where projects are rejected by vote, as what worries him the most. He said that every data center rejected means a victory for China.
Tom also discussed the opposition to data centers, so they both acknowledge the issue. Customers may want more capacity, while residents near the site want something else. Tom considered the industry's attempt to set its own rules sensible and long overdue. He said that a rapidly growing industry is always ahead of its own rules.
Ultimately, it must respond to people's concerns if it wants to continue building. The queue of customers does not provide answers to the community's questions. So, Tom is essentially saying that he sees self-regulation as a way to acknowledge these issues and start a dialogue. Dan was much stricter about the industry's public statements.
He claimed that claims about the disappearance of office jobs amid rising electricity bills had caused a PR disaster. And he worried that slowing American growth would force China to slow down too. So his forecast for data centers depends on more than just tech companies' willingness to build new buildings.
The industry also needs to consider the concerns of people who live near them. Demand explains why companies want to build, but it doesn't decide whether they can do it. Robotics takes us beyond selling computing power to what those computations can actually do. Tom suggests that robots could become the third factor in the economy, alongside labor and capital.
He believes that this additional production capacity could provide growth of up to 7% without corresponding inflationary pressure. It depends on whether robots can provide this additional power . So, he describes the potential huge economic benefits of robotics. The exciting prospect is that work that previously cost too much becomes worthwhile.
But the robot must do this job reliably at a price the customer can justify. A demonstration can show what is possible. A useful business needs it to work well enough for ongoing use. This is what will combine engineering with greater demand, greater sales, and a lasting business opportunity. Dan calls physical AI— machines operating in the real world—the golden goose.
He also estimated the probability of a Tesla and SpaceX merger in late 2027 or by the end of 2027 at over 80%. What connects robotics to its broader vision is the potential for computing to be used in many types of businesses. So what will make these huge predictions more compelling as the build-out continues?
Think of the business that buys equipment to run AI services. The first order provides the equipment manufacturer with a sale. A customer returning because the service was useful will also say a lot. There will be a reason to order again, besides simply wanting to get the scarce chips. Here's how Dan's story of scarcity can turn into years of profitable business.
And repeat orders will give investors reason to believe in the durability of these profits. This is the question behind Tom's comparison of Nvidia and Apple. I believe the demand for useful AI is real, and I am optimistic about what these companies can deliver. I wonder if customers will find it useful enough to come back again.
The forecasts provide a quantitative assessment of this possibility. Companies that consistently offer what people want would give this optimism a real basis for moving forward.
Comments 0