Prognosis: The Artificial Intelligence (AI) Bubble is projected to Pop in 2025. Understanding the Reasons.

Prognosis: The Artificial Intelligence (AI) Bubble is projected to Pop in 2025. Understanding the Reasons.

As the curtains draw on 2024 in less than two weeks, Wall Street is poised for another triumphant year. The prestigious Dow Jones Industrial Average, influential S&P 500, and growth-focused Nasdaq Composite have conquered multiple record-breaking closing highs this year.

A myriad of elements have propelled Wall Street's major indexes to unparalleled territories, including exceptional corporate earnings, stock-split fervor, and Donald Trump's November victory. However, nothing stirs more anticipation than the artificial intelligence (AI) revolution.

The long-term market potential for AI is virtually boundless. Armed with AI, software and systems become increasingly adept at performing their tasks and evolve autonomously without human intervention. According to PwC analysts, AI is forecasted to contribute a staggering $15.7 trillion to the global economy by the end of the decade.

In response to this unprecedented opportunity, prominent AI stocks have seen exponential growth.

Nvidia (up 2.99%) has amassed an impressive $2.9 trillion in market value since the commencement of 2023, with its graphics processing units (GPUs) becoming the dominant force in AI-optimized data centers. Last week, AI networking solutions pioneer Broadcom joined the exclusive club of 11 publicly traded companies worldwide to surpass a $1 trillion nominal valuation. Meanwhile, AI-driven data-mining titan Palantir Technologies (-0.58%) is on the brink of registering a mind-boggling 1,000% increase over the past two years.

These are just a sample of Wall Street's prominent tech stocks that have risen in expectation of heightened demand for AI hardware and software transforming the corporate landscape.

However, while Nvidia's and Broadcom's projections have eclipsed even the most ardent analyst forecasts, there are reasons to believe the AI bubble will burst in the coming year.

History has an impeccable record of quelling parabolic rallies in pioneering innovations

Among the factors that could curtail the nearly vertical ascent enjoyed by AI stocks such as Nvidia and Palantir, none looms larger than history. Although history cannot predict timing, it does have an unblemished record of predicting eventual stagnation or decline in market-leading companies that pioneer cutting-edge innovations.

Around three decades ago, the internet started gaining widespread acceptance and forever altered the corporate growth trajectory. However, the full utility of the internet wasn't understood by businesses for several years, resulting in the formation of the dot-com bubble.

Since then, we have witnessed numerous pioneering technologies, innovations, and trends, including genome sequencing, 3D printing, blockchain technology, cannabis, and the metaverse. The issue is that they all experienced an early-stage bubble-bursting event.

Without fail, both professional and novice investors have consistently underestimated how quickly a new technology or innovation would be adopted and utilized. This eventually leads to disappointment, causing the market leaders of these pioneering trends to lose 80% to 99% of their value.

To clarify, I am not suggesting that AI cannot be a transformative technology. What I am saying is that all new technologies require time to mature. The simple fact that most businesses cannot lay out a clear blueprint as to how they will employ AI to achieve a positive return on investment is a strong indicator that we are in a bubble.

Resolving GPU scarcity would dampen investor enthusiasm (and margins)

Another reason the AI bubble may burst in 2025 is due to the anticipated resolution of GPU scarcity that has catapulted Nvidia's stock to stratospheric heights.

Demand for Nvidia's hardware has been extraordinary, with orders for its H100 GPU and its successor, the Blackwell GPU, significantly backlogged. When the demand for a good or service significantly outstrips its supply, it's typical for its price to increase. Earlier this year, Nvidia was charging around $40,000 for its Hopper chip, which represents up to a 300% premium compared to what Advanced Micro Devices was earning for its Insight MI300X GPUs.

In essence, Nvidia has been leveraging AI-GPU scarcity to raise the price of its hardware and boost its gross margin to the mid-70% range.

However, I am confident that this scarcity advantage will fade in the new year. AMD is ramping up chip production and recently introduced its next-generation MI325X GPU.

Furthermore, many of Nvidia's top customers by net sales are developing their own AI GPUs for use in their data centers. Although Nvidia's chips will remain superior in terms of computing power, these in-house GPUs will be significantly cheaper and easily accessible. This could result in Nvidia losing valuable market share in data centers and experiencing a decline in both pricing power and margins.

The U.S. government could put a halt to the AI rally

Beyond history not being on the side of the AI revolution, the AI rally could also be derailed by actions taken by U.S. regulators.

In 2022 and 2023, regulators under the Biden administration announced restrictions on exports of high-performance AI chips and chip-related manufacturing equipment to China. This affects leading hardware manufacturers such as Nvidia, as well as companies providing equipment for AI solutions. For instance, semiconductor wafer fabrication equipment company Lam Research generated 37% of its revenue from China during the September-ended quarter, and 39% in the quarter prior to that.

With President-Elect Donald Trump set to take office, it's not likely we'll see any relaxation or elimination of these restrictions. Trump has been firm in his approach towards China, the world's second-largest economy, during his first term, and this tough stance is expected to continue.

Moreover, Trump has hinted at imposing a 35% tariff on Chinese imports into the U.S. from Day One. This decision could potentially ignite a trade war, straining relations between the two largest economies and negatively impacting AI product sales to China.

The sky-high valuations of AI stocks are a problem that can no longer be ignored

The main reason for the AI bubble bursting in 2025 is the exorbitant valuation premiums placed on top-performing AI stocks in the market.

Over the last 3 decades, companies at the forefront of cutting-edge innovations have typically hit a peak multiple of 30 to 40 times their trailing-12-month sales. This was the case for companies like Amazon and Cisco Systems before the dot-com bubble burst.

In 2024, Nvidia reached a price-to-sales ratio (P/S ratio) exceeding 40, while Palantir Technologies currently hovers around a P/S ratio of almost 69. While it's impossible to predict when investor enthusiasm will deflate, history has consistently shown that valuations of this magnitude aren't sustainable over the long term.

Although companies with substantial competitive advantages, such as Nvidia and Palantir, deserve a premium valuation compared to their peers, the current P/S ratios of 29 for Nvidia and nearly 69 for Palantir don't add up.

Investors are actively looking for opportunities to invest in AI-related stocks, given the potential for significant growth in this sector. For instance, AI-driven data-mining titan Palantir Technologies has seen its value increase significantly over the past two years.

However, some analysts raise concerns about the current valuations of these AI stocks. For example, Palantir Technologies currently has a price-to-sales ratio (P/S ratio) of nearly 69, which is quite high and could potentially be unsustainable in the long term.

Read also: