Tom Yeung shares an intriguing perspective on the concept of smart money.
My experience growing up in Boston has shaped my understanding of some local customs.
Every significant snowfall seems to bring an odd assortment of items to the curb.
You might spot folding chairs, tables, ironing boards, and occasionally a cooler filled with free beer. It’s all in the name of saving parking spots for those who have just dug them out.
Such is the dedication to preserving parking spaces that the city’s beloved mayor, Thomas Menino, formalized it into law back in 2005. If you’ve just spent a good deal of time clearing snow from a parking spot, you can legally lay claim to it with your household items.
Contrast this with other cities where similar practices aren’t embraced. Take Pittsburgh, where the so-called “parking chairs” face local myths about removal. In Chicago, the entire idea of “dibs” is against the law.
This concept mirrors how businesses operate, too. Unless a company holds a patent or copyright, or has a strong legal team, it’s all too easy for a new competitor to appropriate another’s published work and pass it off as their own.
Think back to the late 1990s, during the internet boom.
The trailblazers of this technological shift weren’t necessarily the ones who built the web’s framework. Remember Lucent? It once dominated telecom in the ’90s but is now a faint memory. The AT&T spinoff poured in $27 billion but subsequently lost $28 billion, overextending its supply.
Lucent, in effect, dug up a place in the internet’s “parking lot”… only to see others claim it.
As it turned out, the real winners of the internet era were companies like Amazon.com Inc. (AMZN) and Netflix Inc. (NFLX), who capitalized on the unmonetized networks around them. Their ongoing success owes much to infrastructure they didn’t have to build.
The same could well apply to firms in the AI infrastructure arena.
Smart investment strategies today highlight that the major returns won’t necessarily come from those pouring the most money into projects. Instead, the savvy winners will be those who effectively integrate AI to streamline their processes and boost profits.
Now, let’s delve further into this.
Consider selling that AI division right away…
AI data centers carry risks similar to older tech.
They require a substantial investment to construct, and just like that, the allure fades.
AI computing primarily leans on services that, at a glance, seem interchangeable. They’re all utilizing the same Nvidia Inc. (NVDA) chips operating on a unified platform. Thus, choosing a provider often boils down to cost.
That’s why Eric cautioned subscribers of the fly investment report to dispose of their Oracle Corporation (ORCL) shares last November when prices hovered around $200. AI data center firms were gearing up to spend billions on expanding infrastructure.
Current long-term profit projections look as flimsy as, say, an ironing board set up on a snowy Chicago street.
Let’s unpack this with an example.
Last September, Oracle struck a massive $300 billion deal to provide OpenAI with 4.5 gigawatts of cloud computing from 2027 to 2032.
Each gigawatt of data center capacity is estimated to cost around $50 billion to build, $35 billion for the required Nvidia chips, and another $15 billion for various additional expenses—a fact confirmed by Nvidia’s CEO, Jensen Huang.
This means Oracle could be on the hook for about $225 billion by 2027 to build these data centers, while generating $300 billion in revenue.
Assuming all proceeds smoothly, Oracle could stand to make a $75 billion profit (that’s $300 billion in revenue minus $225 billion in costs). Plus, there’s a likely chance accountants might allow Oracle to factor in the residual value of the data centers, potentially boosting profits by an additional $70 billion.
This alludes to a remarkable return of $145 billion on a $225 billion investment over eight years.
But that’s a big **if.**
Prepare for a potential downturn
Let’s flip the scenario. What happens if OpenAI misses that contract deadline?
So what?
Oracle’s chances could quickly worsen. Competing data center firms would have to slash prices to lure in other clients (after all, if OpenAI can’t meet the $300 billion bill, who will?). This would lead to plummeting profits.
In a dire outcome, Oracle might end up selling its cloud division at a loss just to satisfy creditors. The stock could sink into the low $100s or worse.
This isn’t merely a scenario pulled from thin air.
What I’m illustrating parallels the events of the dot-com crash.
When the bubble burst, numerous startups couldn’t afford the network equipment they had ordered. That compelled companies like Lucent to provide vendor financing, lending cash to clients, all while lowering prices. Even solvent companies began to hesitate, seeking better deals, squeezing those internet infrastructure players even more.
Currently, Oracle’s value has dropped by 50% since its 2025 peak. The downside risk is clear. Even if pouring optimistic figures into a forecast might suggest it’s undervalued (assuming OpenAI adheres to its $300 billion commitment).
And yet, there are other data center companies like Amazon, Equinix Inc. (EQIX), and Digital Realty Trust Co., Ltd. (DLR), which seem to be more focused on selling rather than acquiring within the AI infrastructure sphere at present.
Conversely, several promising entities are ready to soar in the wake of AI infrastructure.
Invest in this AI sector stock instead
These companies haven’t spent a single cent on data center construction but stand to gain significantly from the resulting computing capacity.
We refer to these as “AI appliances.”
AI appliances strategically leverage AI technology to enhance efficiency, productivity, and profitability. Their capability to integrate AI into existing frameworks could greatly elevate revenues and profit margins.
Just last week, we uncovered one of these AI appliances.
PayPal Holdings, Inc. (PYPL)
PayPal’s approach involves weaving AI into its core financial services. For instance, their own AI-driven fraud detection system continues to evolve as more AI-centric data centers come online.
This payments platform generates impressive amounts of training data daily (about 500 data points per transaction) and is currently preventing around $500 million in fraud each quarter. In 2024, reports showed that PayPal utilized over 2,000 database servers to execute one million queries per second, or about a trillion service requests each day. The cost-effective computing power provided by Oracle will further enhance PayPal’s AI applications.
It’s crucial to note, however, that not all AI appliances will thrive. Just because a parking spot is cleared, it doesn’t mean someone can easily take it.
In fact, there often seems to be room for just one standout.
The real challenge lies in identifying which AI appliances will harness the existing infrastructure to their advantage—and which will falter.
Choose a winner
Eric categorizes AI appliances among four distinct AI investment types (the others being builders, enablers, and survivors).
He has wisely included successful AI appliance firms in his fly investment report portfolio.
For instance, one of his picks in the healthcare AI appliance sector (represented by the blue line) has surged by 30% over the past three months, even as Oracle’s stock (shown in white) has declined sharply during that time.
This particular company is utilizing AI to innovate a brand new drug, poised for major success in the coming years.
We think this is only the beginning.
In the upcoming issue of the fly investment report, Eric will delve deeper into AI appliance strategies related to logistics, robotics, and even online dating, all of which will reap rewards from the growing landscape of cost-effective AI computing power. These enterprises are biding their time while major firms like Amazon and Oracle invest heavily in building out data centers.
Eric offers further insights into this AI category in the February edition, expected to be published soon. In this issue, we will also showcase new selections that have been added to our portfolio.
Be sure to check how you can get involved with the fly investment report today.
And keep an eye on your inbox.
Until next time,
Thomas Yang, CFA
market analyst, investor place
