It pays to note that much of Big Tech – Meta, Alphabet, Microsoft and Amazon – has price-to-earnings multiples of around the 20s. — The Straits Times
SINGAPORE investors have heard it before – warnings that a bubble is about to burst.
The hypothetical artificial intelligence (AI) bubble, tied to a rise in stock prices of AI-related firms, eager investors and feverish enthusiasm for data centres, has been compared with the 1999 dotCom days, when hype outran real profits.
It started in late 2022 with the overnight fame of OpenAI’s ChatGPT, and culminated in a year of frenzied deal-making in 2025: Firms such as Oracle, Nvidia, AMD and Amazon each pledged billions of dollars to OpenAI in anticipation of orders to fulfil the startup’s yet-to-materialise demand.
But here is the problem: OpenAI expects to burn through billions of dollars before it turns in a profit – US$115bil until 2029, in fact – before cash comes in, according to The Information in September. The tech news outlet did not cite a source.
These are all classic signs of froth, say some. But there are good arguments on why it is not. Here’s a list.
> AI is a narrative that is playing out.
Bears are worried that the world will be a glut of data centres when enterprises find insufficient compulsion to pay for this shiny new technology.
But people forget that the world is made up of parts other than enterprises that dare sit out what their competitors might effectively deploy.
Governments are pushing for connectivity, plans are under way for robots and autonomous buses and cars, and consumers are demanding more services and goods on the go, as well as lots of videos, songs and games.
All that needs compute infrastructure.
In its report on the state of AI in 2025, consulting firm McKinsey wrote that while organisations have yet to show bottom line lifts from AI and many have yet to integrate the technology, 68% of them are piloting, scaling or have fully scaled AI.
So, enterprise use of AI is moving.
In real life, marketers are using the technology to propagate content and drive round-the-clock programmatic media buys and generate personalised leads.
Doctors are pointing to AI boosts in diagnostic imaging, predictive analytics and ops automation to reduce clinician burnout and wait times.
In a note dated Dec 11, Goldman Sachs Research analysts forecast that AI’s slice of the overall data centre market will double to 30% over the next two years, taking share from conventional and cloud workloads.
The investment bank’s senior equity analyst Jim Schneider said: “We believe (data centre) occupancy will continue to tighten through the medium term before the market loosens.”
In short, demand for data centres in 2026 is here to stay.
> Easy money.
OpenAI sits squarely at the centre of this narrative – a private valuation built more on potential, surrounded by governance drama and outsized expectations.
Over the course of 2025, tech firms including Oracle Cloud Infrastructure and chipmakers Nvidia and AMD did not bat an eyelid to commit an eye-watering total of US$1.3 trillion in forward-looking deals with the startup.
“Circular deals” critics pointed to OpenAI backers pumping billions of dollars into the startup with the expectation that it will, in turn, pour back the funds to buy their chips, clouds and data centres as making up “artificial revenues”.
Analysts point to low interest rates and the coffers of these cash-rich technology giants, which have flooded AI startups and infrastructure projects with capital.
In their upcoming financial year, Meta, Alphabet, Microsoft and Amazon are collectively expected to fork out more than US$400bil on capital expenditure, primarily in AI infrastructure.
But these companies have strong war chests.
In their last financial year, they generated a total of more than US$300bil in net profits, and held over US$200bil in combined cash reserves.
Nvidia, for one, ended its fiscal 2025 with more than US$43bil in cash, cash equivalents and marketable securities, and essentially no net debt. That is a large cushion against any single customer’s hard landing.
Talk of interest rates rising has somehow soothed.
In December, the US Federal Reserve lowered the target range for the federal funds interest rate by 25 basis points to 3.5% to 3.75%, marking its third rate cut in 2025.
US President Donald Trump has indicated that his nominee to replace Fed chair Jerome Powell in 2026 will be someone who believes in lower interest rates “by a lot”.
Boosting growth
Americans go to mid-term polls in November 2026, and observers expect the President to press for a looser monetary policy to boost growth.
So the bet is: Money will keep rolling.
> Valuations have not gone mad.
Nvidia spent much of 2025 trading on price-to-earnings (PE) multiples in the high 40s to 50s, well above both the S&P 500 average and its own pre-AI-boom history.
If that is not AI exuberance, what is?
But it pays to note that much of Big Tech sits at less extreme levels. As of December, Meta, Alphabet, Microsoft and Amazon hover mostly from the 20s to mid-30s in PE multiples.
As of late 2025, the S&P 500’s average PE ratio trailing 12 months has actually been around 30.
Compare this with the dotCom boom in 1999 when firms like Cisco, Intel and Oracle traded at multiples ranging from 40 to as high as over 200.
Microsoft, Meta, Nvidia, Amazon and Alphabet together command about US$16 trillion in combined market capitalisation as at December, representing about 30% of the S&P 500 total value.
Deep diversification
These Big Five firms have established deep diversification, as well as integration, across the entire AI stack, insulating them from single-point failures.
Google and Amazon, for example, are building and at the same time reinforcing their infrastructural cloud layer with their respective models.
Both are also developing their own chips.
The players are dominant in their core businesses – Google in search, Amazon in cloud and eCommerce, Meta in social media and Microsoft in cloud and enterprise tools.
Nvidia’s business is spread across an ecosystem of data centres, networking, software and gaming businesses, making it able to cushion even a hard landing by OpenAI.
So, the stock market is unlikely to be heading for ground zero even if demand for AI slows.
Although the rise of Chinese AI models, robotics and related technology could threaten to rain on the party in 2026, these big tech firms have enough heft to absorb the shocks. — The Straits Times/ANN
Krist Boo is a senior business correspondent of The Straits Times. The views expressed here are the writer’s own.
