The 2002 Treasury yield has returned, but the tech stocks facing it belong to an almost unrecognizable economy.

The 2002 Problem: Interest Rates Are Back. Tech Is Almost Unrecognizable.

Finance by Léa Valmont

The last time the US 10-year Treasury yield was this high, The Lord of the Rings: The Two Towers was still months away from theaters.

Windows XP was Microsoft’s flagship operating system. The original Xbox had barely survived its first Christmas. Apple’s most exciting new product was an iPod with a mechanical scroll wheel and enough storage for roughly 1,000 songs. Facebook did not exist. YouTube did not exist. Tesla did not exist. The iPhone was five years away.

Amazon was still primarily an online retailer trying to prove that surviving the dot-com crash had not been a temporary achievement.

Nvidia sold graphics processors to PC gamers.

Google had not yet gone public.

In October 2026, the US 10-year Treasury yield climbed above 5.3%, reaching its highest level since early 2002. That comparison immediately sounds ominous because financial markets have spent most of the intervening quarter-century becoming accustomed to cheaper money.

But there is a problem with comparing 2026 to 2002.

The interest rate may have returned. The technology industry it is pricing has not.

Twenty-four years ago, investors were sorting through the wreckage of the dot-com bubble and trying to determine which internet companies had actual businesses underneath their promises. Today, many of the world’s dominant technology companies generate tens of billions of dollars in annual profit, own enormous infrastructure networks and sit on balance sheets that would have seemed almost absurd in 2002.

At the same time, artificial intelligence is transforming those supposedly asset-light technology giants into some of the largest builders of physical infrastructure on Earth.

That creates what we might call the 2002 problem: the market has rediscovered a cost of capital from another technological era, but the companies exposed to that cost have changed beyond recognition.

Understanding what a 5.3% Treasury yield means for technology stocks therefore requires more than finding the last time the number appeared on a chart.

You have to travel back to the world that existed around it.

Table of contents

The 10-year Treasury yield just traveled back to 2002

On October 1, 2026, the yield on the benchmark US 10-year Treasury briefly climbed to roughly 5.34%, its highest level since early 2002 and above the peak reached before the 2008 financial crisis.

That matters far beyond the bond market.

The 10-year Treasury sits near the center of global finance. Its yield influences mortgage rates, corporate borrowing costs and the discount rates investors use when deciding what future profits are worth today. When it rises sharply, almost every financial asset has to negotiate with a more demanding alternative.

A Treasury security yielding more than 5% gives investors something they barely encountered during much of the post-financial-crisis era: a relatively high nominal return from an asset backed by the US government.

Stocks still offer something bonds cannot — ownership in businesses whose earnings can grow — but that future growth now has to compete against a much higher starting yield.

For technology companies, that competition can become particularly uncomfortable because so much of their valuation often depends on profits expected years into the future.

Yet simply saying “rates are back at 2002 levels” risks creating a false historical analogy.

Because 2002 was a very strange place.

Technology in 2002 belonged to another civilization

By 2002, the dot-com dream had collided violently with accounting.

The Nasdaq had collapsed from its March 2000 peak. Internet companies had disappeared. Telecommunications businesses had imploded under debt accumulated during the infrastructure boom. Investors who had spent the late 1990s paying extraordinary prices for anything connected to the web were discovering that technological importance and shareholder returns were not the same thing.

The survivors inhabited a digital economy that now looks almost primitive.

PCs were still the center of personal computing. Broadband adoption was expanding, but dial-up internet remained part of everyday life. Software arrived in boxes or through corporate licensing agreements. Streaming video was not a meaningful consumer market. Cloud computing had not yet become an industry. Social media as we understand it barely existed.

Apple had launched the first iPod only the previous year.

Amazon Web Services was years away.

Facebook would not appear until 2004.

YouTube would follow in 2005.

The iPhone would arrive in 2007.

Bitcoin was still an idea waiting for someone to have it.

And artificial intelligence, while very real as an academic and engineering discipline, had nothing resembling its current position at the center of corporate investment.

Even familiar company names represented very different businesses.

Amazon was selling books, electronics and other goods online rather than renting computing infrastructure to much of the internet. Nvidia was known primarily for graphics hardware. Microsoft revolved around Windows and Office. Google was a rapidly growing private search company rather than one of the world’s largest advertising and cloud platforms.

Looking at a 2002 interest rate without looking at that 2002 economy is like comparing two video games because they happen to display the same frame rate.

The number matches.

Almost everything underneath it has changed.

Microsoft shows how enormous the transformation has been

Microsoft offers perhaps the cleanest bridge between the two eras because it was already a technology giant when the 10-year Treasury last occupied this territory.

In fiscal 2002, Microsoft generated $28.4 billion in revenue and $7.8 billion in net income. It ended the year with approximately $38.7 billion in cash and short-term investments.

Those were formidable numbers.

The company’s business, however, belonged unmistakably to the PC age. Microsoft described strong demand for Windows XP as one of the forces driving its performance. The original Xbox was a new consumer business. Enterprise software meant products such as SQL Server and the company’s .NET server family.

The economics were equally revealing.

Microsoft generated $14.5 billion in operating cash flow during fiscal 2002 while spending only about $770 million on additions to property and equipment.

That is an extraordinary reminder of what the classic software model looked like.

Microsoft could create enormous economic value without constantly pouring tens of billions of dollars into physical infrastructure. Software was expensive to develop but extremely cheap to reproduce. Once Windows or Office existed, distributing another copy required very little incremental capital.

That basic economics helped define the mythology of technology investing for decades: scalable businesses, high margins, limited physical assets and enormous operating leverage.

Artificial intelligence is now complicating that mythology.

Dramatically.

Why does the 10-year yield matter to tech stocks?

A company is theoretically worth the present value of the cash it can generate for its owners in the future.

The word present is where interest rates enter the equation.

Money expected ten years from now is worth less than money received today, so investors discount future cash flows. When interest rates rise, that discount becomes more severe.

Imagine two companies.

The first generates enormous cash flows today but grows relatively slowly. The second earns little now but could generate spectacular profits a decade from today.

Higher discount rates generally hurt the second story more because a greater proportion of its assumed value exists far in the future.

That is one reason rapidly growing technology stocks became so sensitive to interest rates during the ultra-low-rate era. When safe yields approached zero, distant growth became extraordinarily valuable. Investors had fewer attractive alternatives, and the mathematical penalty applied to future earnings was relatively small.

A 5% Treasury world changes the equation.

Future profits have to be discounted more aggressively, while investors can earn meaningful yields without taking equity risk.

This does not mean technology stocks automatically fall whenever Treasury yields rise. Earnings growth, margins, competitive position, inflation expectations and countless other variables matter simultaneously.

It means the hurdle becomes higher.

Today’s Big Tech companies are much harder to kill with rates

Here the comparison with 2002 becomes interesting rather than merely nostalgic.

The modern technology sector contains speculative companies whose valuations depend heavily on future growth, just as previous technology cycles did. But its largest companies are not generally fragile dot-com startups waiting for a business model to arrive.

They are among the most profitable corporations ever created.

Microsoft, Alphabet, Meta, Amazon and other technology giants operate businesses used by billions of people and millions of companies. They dominate search, cloud computing, digital advertising, enterprise software, e-commerce, social networks and mobile ecosystems.

That gives the current market a fundamentally different foundation from the most speculative corners of the dot-com era.

A high Treasury yield can reduce the valuation investors are willing to assign those earnings.

It does not make the earnings disappear.

Large cash balances and powerful operating cash flows can also reduce direct dependence on external financing. A profitable hyperscaler funding investment internally experiences higher rates differently from a young technology company that must repeatedly issue debt or equity simply to continue expanding.

This distinction is easy to lose when people talk about “tech” as though it were one homogeneous trade.

At 5.3%, the 10-year Treasury is not applying the same pressure to every company with an AI slide in its investor presentation.

Balance sheets matter again.

Cash flow matters again.

The distance between technological promise and economic self-sufficiency matters enormously.

Then AI turned Big Tech into heavy industry

This is where the historical comparison takes another strange turn.

The largest technology companies are vastly stronger financially than they were in 2002.

They are also becoming much more capital intensive.

S&P Global Ratings estimated in May that five major cloud providers it follows — Alphabet, Amazon, Meta, Microsoft and Oracle — could collectively spend approximately $750 billion on capital expenditures in 2026, equivalent to roughly 38% of their combined revenue.

The principal reason is AI infrastructure.

Training and serving increasingly capable models requires GPUs and other accelerators, networking equipment, cooling systems, enormous data centers and extraordinary quantities of electricity. Securing enough computing capacity has become a strategic objective important enough to justify investment programs normally associated with physical infrastructure industries.

Meta alone has indicated capital expenditures could reach as much as $145 billion in 2026.

The technology sector has spent decades moving away from physical constraints. Cloud computing abstracted servers. Software subscriptions abstracted installation. Smartphones compressed cameras, maps, music players and computers into a single object.

AI is pushing in the opposite direction.

Behind every seemingly weightless chatbot sits an industrial system.

Semiconductor fabrication plants have to manufacture the chips. Data centers have to house them. Electrical grids have to power them. Cooling infrastructure has to remove the heat. Fiber networks have to connect everything. Land has to be acquired, facilities constructed and equipment financed.

The biggest technology companies are still software businesses.

They are increasingly software businesses with industrial-scale construction budgets.

That makes today’s high-rate environment different from both the dot-com era and the zero-rate technology boom that followed the financial crisis.

The real pressure point may be valuation

A company does not need to borrow heavily for higher Treasury yields to matter.

The pressure can arrive through valuation.

When the risk-free rate rises, investors can demand higher expected returns from stocks. That can translate into lower multiples even when corporate profits remain healthy.

This is particularly relevant to AI because today’s enormous infrastructure spending is justified by profits that may take years to fully materialize.

The largest hyperscalers can afford the buildout.

That does not automatically mean every dollar invested will produce an attractive return.

A data center is not valuable merely because it contains expensive GPUs. Its economic value depends on how much profitable demand eventually runs through those machines, how quickly hardware becomes obsolete, what competitors charge for equivalent computing power and whether AI products generate enough revenue to justify the infrastructure supporting them.

At low interest rates, investors can be patient with distant payoffs.

At 5.3%, patience has an opportunity cost.

Every additional year required for an AI investment to produce meaningful cash flow competes with a Treasury security already producing more than 5% nominally.

That does not destroy the AI investment thesis.

It changes the price of waiting for it to work.

This is not the dot-com bubble, which does not make it safe

Whenever technology valuations collide with rising interest rates, 2000 inevitably enters the conversation.

The analogy is useful up to a point.

Both periods feature extraordinary enthusiasm surrounding a technology capable of transforming the economy. Both produced aggressive investment in infrastructure. Both encouraged investors to project enormous future markets from technologies whose eventual commercial structure remained uncertain.

But the differences are substantial.

Many companies at the center of today’s AI boom already generate huge revenues and profits from businesses that exist independently of AI. Microsoft does not need Copilot to prove that Microsoft Office has customers. Alphabet does not need Gemini to establish that digital advertising exists. Amazon does not need generative AI to demonstrate demand for cloud infrastructure.

The speculative question has moved.

In 2000, investors frequently had to ask whether an internet company would ever develop a viable business.

In 2026, the more relevant question for the largest technology companies is whether the enormous incremental capital devoted to AI will earn returns high enough to justify its cost.

That is a healthier starting point.

It is not a trivial one.

An established company can destroy enormous amounts of shareholder value by investing too much capital at inadequate returns. Financial strength increases the amount a company can spend before discovering that the economics were wrong.

The hyperscalers have something the dot-com startups rarely possessed: the ability to finance the experiment themselves.

That also means they can make the experiment unimaginably large.

Same yield, two completely different technological worlds

There is something wonderfully absurd about the 10-year Treasury returning to 2002 levels.

The number has completed a round trip through technological history.

During the journey, the desktop internet became the mobile internet. Apple moved from the iPod to the iPhone and then to custom silicon. Amazon evolved from an online retailer into a cloud infrastructure giant. Google went from private search engine to one of the world’s dominant corporations. Nvidia moved from gaming graphics to the center of the AI economy.

Social networks appeared, conquered the internet and reshaped politics and culture.

Streaming replaced much of physical media.

Cloud computing turned servers into rentable infrastructure.

Bitcoin created an entirely new asset class.

Generative AI made conversation itself a computing interface.

And after all that, investors once again find themselves staring at a 10-year Treasury yield beginning with a five.

The temptation is to conclude that high rates must produce the same outcome they did in the previous era.

History is rarely that cooperative.

The technology companies facing today’s bond market are richer, larger and more profitable than their predecessors. Their products are embedded far deeper into the global economy. Many can fund extraordinary investments from their own cash flows rather than depending entirely on capital markets.

But they are also attempting an infrastructure buildout whose scale would have been almost incomprehensible to the software industry of 2002.

That combination is what makes the current moment unusual.

The 2002 technology industry was recovering from discovering that digital dreams still required profits.

The 2026 technology industry already has the profits.

Now it has to prove that spending hundreds of billions of dollars on the next digital dream will create enough more of them.

The 10-year Treasury has returned to where it was 24 years ago.

Tech has spent those 24 years changing almost everything else.

The 2002 Problem: key questions

How high is the 10-year Treasury yield?

The US 10-year Treasury yield briefly reached roughly 5.34% on October 1, 2026, its highest level since early 2002. Yields have remained above 5% in early October.

Why does the 10-year Treasury yield matter for tech stocks?

Treasury yields influence the discount rates investors use to value future corporate cash flows. Higher rates can reduce the present value assigned to distant earnings and also provide investors with a more attractive low-risk alternative to equities.

Why compare 2026 with 2002?

The 10-year Treasury has returned to a yield last seen in early 2002, creating a useful comparison between two radically different technological eras. In 2002, the industry was recovering from the dot-com crash and remained centered on PCs, packaged software and the early internet. In 2026, cloud computing, smartphones and AI dominate the landscape.

Were technology companies profitable in 2002?

Some certainly were. Microsoft, for example, generated $28.4 billion in fiscal 2002 revenue and $7.8 billion in net income. The dot-com crash should not be interpreted as evidence that the entire technology sector lacked profitable businesses.

Why is AI making Big Tech more capital intensive?

AI requires enormous physical infrastructure, including advanced chips, networking equipment, data centers, cooling and electricity. S&P Global Ratings estimated that five major cloud providers it rates could collectively spend about $750 billion on capital expenditures in 2026.

Does a 5% Treasury yield mean tech stocks must fall?

No. Interest rates are only one variable affecting equity valuations. Earnings growth, margins, competitive advantages, inflation, economic conditions and investor expectations also matter. Higher yields simply raise the return available from government bonds and can increase the valuation hurdle faced by equities.

Is 2026 another dot-com bubble?

There are similarities, particularly around technological enthusiasm and large infrastructure investments, but major differences make a direct comparison misleading. Many of today’s largest AI investors already operate highly profitable global businesses. The central question is increasingly whether their enormous AI investments will generate adequate returns rather than whether they possess viable businesses at all.

Is this financial advice?

No. This article analyzes interest rates, technology companies and historical market conditions. It does not recommend buying, selling or holding any security.

Sources: