Under the Spotlight: Nvidia Corporation ($NVDA)

ICYMI: Do your own research and make your own decisions. This article drills down on a specific company, however, it is not a recommendation to invest in the company and should not be taken as financial advice. Got a stock you want covered? Tell us here.

The last time we covered Nvidia ($NVDA) in Under The Spotlight, ChatGPT wasn’t yet on anyone’s radar. It was 2021 and the world was a different place.

Back then, Nvidia was largely a gaming chip company that had started venturing into cryptocurrency mining. It had a market value of US$500B and we called it a ‘company hidden in plain sight’.

Those days are well behind it. Today, Nvidia is the world’s most valuable public company and the first to reach a US$4T market cap this year. It’s so all-encompassing it now makes up around 8% of the S&P 500.

To put that growth into perspective, if you’d invested $1,000 into $NVDA five years ago, that stake would be worth roughly $13,000 today. That’s a share price gain of more than 1,200%.

The signs were there if you knew where to look. CEO and founder Jensen Huang was already talking up Nvidia’s push into artificial intelligence back in 2020, with a vision to become the ‘leading computing company for the age of AI’. Credit where it’s due – he delivered.

But with new competition emerging and an increasingly complex web of circular deals between AI and mega-tech firms, investor confidence has begun to wobble.

Godfather of AI

Like other chipmakers now riding the AI wave, Nvidia started off making chips for other purposes, namely gaming and computing. While its graphics chips (GPUs) were competitive  before, they’ve found a near monopoly in AI infrastructure. 

GPUs are essential for the data centres used to train AI models, and Nvidia’s are the best in the business. Data centre revenue in Q3 hit a record US$51.2B, up 66% year on year. Compare that with its closest rival AMD ($AMD), which reported US$7.7B in total revenue across all segments in the June quarter, and Nvidia’s dominance comes into focus.

Nvidia’s edge isn’t just about hardware. Much of its pricing power – and its enviable 73.6% adjusted gross margin – comes from CUDA, its proprietary computing platform. 

CUDA lets software talk directly to GPUs, effectively turning them into a single, coordinated compute engine. Most modern AI models and workflows are built around this ecosystem, which makes it costly and complex for customers to switch away, even when alternative chips come cheaper. 

But now a new and emerging threat is developing on its doorstep. Several of Nvidia’s largest customers, including Alphabet and Amazon, are aggressively developing their own AI hardware. In November, reports that Meta Platforms ($META) was planning to use Google’s ($GOOGL) custom-built AI chips, known as TPUs, in its data centres sent Nvidia shares down 2.6%.

Google’s TPUs are designed by Broadcom ($AVGO), which has also struck billion-dollar custom chip deals with OpenAI and rival Anthropic in recent weeks. These chips can be tailored to specific workloads, potentially making them faster, cheaper and more energy efficient.

Soon after, Amazon unveiled its new AI chip, Trainium3, claiming it was four times faster than previous models. Anthropic already has a deal in place to use millions of Amazon’s custom chips by the end of 2025. OpenAI is also reportedly in talks with Amazon over a US$10B chip deal, according to The Information.

AI boom or bust?

The increasingly tangled relationships between the world’s biggest tech firms are fuelling concerns that the AI boom could be drifting into bubble territory.

Nvidia’s partnership with OpenAI, announced in September, outlined plans for Nvidia to sell 10GW of chips to OpenAI for cash, while also investing up to US$100B in OpenAI equity. Another deal will see Anthropic buy US$30B of Nvidia-powered Microsoft Azure capacity, while Microsoft and Nvidia will jointly invest US$15B into Anthropic.

These deals underscore Nvidia’s importance, but they also blur the lines between customer, partner and investor in ways markets are still trying to price.

On the surface, Nvidia’s blockbuster earnings in November should have calmed nerves. The company reported a record US$57B in Q3 revenue, up 62% from a year earlier. Huang described demand for its Blackwell chips as ‘off the charts’.

Instead, $NVDA shares are down around 17% since the start of November. Other AI-linked names have seen similar pullbacks. Broadcom and Oracle ($ORCL) have both sold off in recent days despite record earnings, while the PHLX Semiconductor Index is down 5%.

The message from investors is shifting. They’re no longer backing AI projects on faith alone. Now they want to see whether the numbers stack up.

Most large-scale AI projects are still years away from breaking even. UBS estimates AI capital expenditure will reach US$423B in 2025. OpenAI reportedly needs around US$1.4T over the next eight years to maintain the infrastructure behind ChatGPT, despite generating just US$13B in revenue this year.

A recent UBS survey found fewer than 20% of IT firms are using AI at scale, with unclear return on investment cited as the biggest hurdle. Analyst Karl Kierstead called it a reminder for tech investors to ‘remain sober’ about expectations for AI-driven revenue growth in 2026.

Still, it matters who’s footing the bill. Microsoft, Amazon, Alphabet and Meta generate hundreds of billions of dollars in annual cash flow, giving them the firepower to fund AI investment for years, even if returns take time to materialise.

Valuation check

By traditional measures, Nvidia looks expensive. The stock trades at around 44 times earnings (trailing P/E), compared with the S&P 500’s 10-year average of 18.7 and its current multiple of roughly 23 times.

But price-to-earnings ratios are a blunt tool for fast-growing companies. They reflect past earnings, not how quickly profits are expanding or what lies ahead. That’s why comparisons with the dot-com era can be misleading. Many tech stocks back then were valued on promise alone, with little clarity on when profits would arrive.

Nvidia’s case is different. Based on analysts’ 2026 forecasts, its forward P/E drops below 25 times, reflecting expectations of rapid earnings growth. The company also holds around US$61B in net cash and securities and generated roughly US$22B in free cash flow in Q3.

Of the 28 analyst ratings featured on Stake, 27 rate Nvidia a ‘buy’, with an average price target of US$263 as of 18 December 2025. That implies an upside of around 52% – an exceptionally bullish outlook by any standard.

Is $NVDA a buy or sell?

Nvidia sits at the centre of the most powerful investment theme of this decade, and that’s exactly what makes it so divisive.

On one hand, it’s delivering profits at a scale few companies in history have matched, with demand for its chips still exceeding supply and analysts overwhelmingly backing further upside. On the other, the conversation has clearly shifted. Investors are no longer debating whether AI will change the world, but how long the spending boom can last, who ultimately earns a return and how much of Nvidia’s future success is already priced in.

Nvidia can still be an outstanding business even if its share price remains volatile. For investors, the real question isn’t whether Nvidia wins the AI race. It’s whether today’s valuation leaves enough room for disappointment in a market that’s starting to demand proof, not promises.

This is not financial advice nor a recommendation to invest in the securities listed. The information presented is intended to be of a factual nature only. Past performance is not a reliable indicator of future performance. The author of this article and other employees of Stakeshop Pty Ltd may hold positions or have financial interests in the company (or companies) discussed above. As always, do your own research and consider seeking financial, legal and taxation advice before investing.