I'm LongbridgeAI, I can summarize articles.The article recommends a diversified $5,000 investment strategy across three AI infrastructure layers: Amazon (AMZN), Taiwan Semiconductor Manufacturing (TSM), and Applied Digital (APLD). Amazon serves as the anchor with strong AWS growth and profitability. TSMC offers exposure to essential chip manufacturing with high margins. Applied Digital provides speculative upside through data center construction despite current losses. The suggested allocation is $2,500 in Amazon, $1,750 in TSMC, and $750 in Applied Digital.
The artificial intelligence (AI) build-out isn't a single trade. The money flows through layers: Cloud platforms rent out computing capacity, a foundry manufactures the chips underneath it, and data center builders put up the buildings that house all of it.
That means an investor with $5,000 to put to work doesn't have to bet on one layer. I'd split the money across three stocks -- Amazon (AMZN +1.94%), Taiwan Semiconductor Manufacturing (TSM +1.22%), and a small slice of Applied Digital (APLD +2.48%). Each contributes something the other two can't.
And the sizing matters as much as the names.
Image source: The Motley Fool.
Amazon: the anchor
Amazon gets the biggest slice because it pairs a fast-growing AI business with the profits to pay for it.
Amazon's cloud computing segment, Amazon Web Services (AWS), hasn't grown this fast in 18 quarters. Second-quarter revenue rose 37% year over year, reaching $42.2 billion, and the segment turned that growth into $16.6 billion of operating income, up about 64%. In other words, the cloud layer of the build-out is accelerating -- and profitably.
Even so, AWS is only about a fifth of Amazon's revenue. The rest comes mostly from retail and advertising, which gives the stock a footing the pure AI plays lack.
The spending is heavy, to be sure. Free cash flow ran about $7.6 billion negative over the trailing 12 months, mostly because of the company's AI infrastructure investments. But that kind of spending is exactly what the other two stocks in this basket get paid from.
Shares trade at about 24 times forecast earnings for next year, the basket's most ordinary valuation on its most diversified business. That seems a reasonable price for an anchor.
NASDAQ: AMZN
Key Data Points
Taiwan Semiconductor: everyone's chipmaker
Whoever wins the AI chip race, chip foundry giant Taiwan Semiconductor manufactures advanced processors for most of the leading designers. Owning the foundry means not having to guess which chip design comes out on top.
Demand has been climbing. Revenue was up 36% year over year in the second quarter, then 45% in July, and then 53% in August, a monthly record of NT$514.8 billion (about $16 billion). Notably, revenue through the first eight months of 2026 is running 39% ahead of the same period last year.
The work is lucrative, too, with a gross margin of 67.7% in the second quarter.
At about $431 as of this writing, shares trade at about 20 times expected earnings for next year -- a lower price-to-earnings multiple than Amazon's, for a faster-growing business. However, the discount arguably reflects concentration. TSMC's growth is tied closely to AI infrastructure spending, and its most advanced factories sit in Taiwan. If the big spenders pull back, the foundry would feel it quickly.
NYSE: TSM
Key Data Points
Applied Digital: the speculative slice
The last slice is the smallest because the business is the least proven. Applied Digital builds AI data centers and rents them to large tenants. Tenants have signed 15-year leases on about 1.4 gigawatts of the company's capacity, contracts that total about $36 billion in future rent. Yet only 175 megawatts of that leased capacity was live when the company reported in late July.
The gap between those two numbers is both the investment case and the risk. Fiscal 2026 revenue (the year ended May 31) was $611.3 million, up 167% from the year before, alongside a net loss of $249.2 million.
Rent begins only as buildings are finished and handed to tenants, so nearly all of the contracted money is still ahead. And construction timelines can slip.
Why include it at all? The leases are take-or-pay agreements (rent is owed even if a tenant ends up not using the capacity). And the company's stock market value is about $7.8 billion, a small fraction of what the finished portfolio is under contract to collect.
If the campuses keep getting delivered on schedule, the stock could be worth far more. That uncertainty is why the position stays small.
NASDAQ: APLD
Key Data Points
How would I split the $5,000?
I'd put $2,500 in Amazon, $1,750 in Taiwan Semiconductor, and the final $750 in Applied Digital.
Half the money goes to the steadiest layer, the company that can fund its AI ambitions from its own profitable businesses. Taiwan Semiconductor gets the next slice, offering faster growth and a cheaper price-to-earnings multiple in exchange for more concentration risk.
Applied Digital gets a sliver, sized small enough that a construction delay wouldn't derail the whole plan.
Of course, all three ride the same underlying trend, and a slowdown in AI infrastructure spending would hit the whole basket -- Amazon least, Applied Digital most. The lopsided split is deliberate.
Ultimately, I'd rather own three layers of the build-out than one company's slice of it. I'd buy all three at these prices, in these proportions.