The AI boom: The Magnificent Seven may no longer be the market leaders. And that’s ok

The AI boom: The Magnificent Seven may no longer be the market leaders. And that’s ok

This content is brought to you by OMBA*
Published on

2025 has been a very ‘eventful’ year so far.

Many market participants entered the year optimistic that the Trump 2.0 administration would herald a business-friendly approach, boosting already strong US asset prices. The reality has been somewhat different, with the tariff trade war front and center and driving substantial volatility. In this context, US equities have been under pressure relative to their counterparts in other parts of the world, while the famed “Magnificent Seven” have underperformed broader US indices. It seems not many people want to mention AI these days.

S&P500 (SPY ETF) relative to MSCI All Country World ex USA index (ACWX ETF): – 13% underperformance from USA YTD

However, when we delve below the surface of index and even sector returns, the action isn’t all over in tech-land. Far from it - there are stocks that have delivered astounding returns year-to-date. 

Looking at year to date returns for the S&P500, as well as the Magnificent Seven index of America’s favourite large tech stocks, all its individual constituents, as well as selected technology / tech-enabled stocks, some notable observations are clear: 

YTD returns: S&P500, MAGS index, other selected technology / tech-enabled stocks (all data through 9th May 2025)

  • Notwithstanding the recent rally in equities, it remains a tough start to the year for equities with the S&P500 still down 3.4% YTD as of Friday, 9th May. This is especially poor for a year with a new US presidential term.

  • The Magnificent Seven index (MAGS; red dashed line) has sharply underperformed this year – down 12% YTD. These are the biggest companies in the world that led the 2023/24 rally in US indices to their highs – indeed, the MAGS index outperformed the S&P500 by 35% in 2024. 

  • Aside from Tesla – the worst performer in this group - whose profits have been hard-hit by volume declines and price-cuts which have thus far failed to stimulate demand, it is notable that Nvidia – the poster child for the AI trade in 2024 - remains 13% below the levels it ended 2024 – despite still-strong earnings growth. Similarly, Apple stands out as a major relative loser, down 21% YTD. Alphabet has had its share of struggles too, down 19%.

  • On the other end of the spectrum, we have some major winners: Palantir Technologies has gained 54%, Duolingo +58%, and Uber +37%, to name just a few. Similarly, some of the cybersecurity stocks have delivered stellar returns, with CrowdStrike (CRWD) gaining 20% YTD.

What do we think is going on here?

This appears to be normal rotation of market leadership that happens as a cycle evolves and matures. In this case, we think the AI revolution remains fully intact but is evolving from its early drivers – the semiconductor infrastructure layer – to the companies that stand to benefit from the implementation of this infrastructure into their business models. Software companies that successfully monetize AI rather than being displaced by it come to mind. 

The first beneficiary of the AI boom was Nvidia – and it makes sense that they were the market leader, as cloud service providers embarked upon a massive capital investment boom to refit their datacenters with the GPUs needed to run their AI services. From November 2022 (when OpenAI’s ChatGPT service was announced to the world), Nvidia’s share price gained 200% through the end of 2023, and then a further 214% by the time it peaked in January this year. Despite its recent rally off the lows, it remains ~25% below peak levels. Furthermore, when eyeballing the chart of its relative returns, relative momentum began to fade since as far back as June 2024 (its initial “blow-off top”). 

Nvidia share price relative to Nasdaq 100 (QQQ)

Nvidia is still expected to grow its profits by over 50% this fiscal year, but the rate of change of profit growth has almost certainly peaked and despite recent earnings reports from companies such as Microsoft and Meta suggesting that AI silicon supply remains constrained, we think it’s more likely than not that the law of large numbers means that the days of 100%+ earnings growth are behind us. This has resulted in Nvidia’s forward P/E multiple compressing from ~45x in mid-2024 to 26x today. Investors are clearly not just worried about this years’ earnings, but what may happen to their cycle over the next 3+ years if AI capex starts to peak – or worse, roll over.

Nvidia forward P/E: trough multiple on (close to) peak earnings? 

Another complication that has crept into the investment case for Nvidia is that the company’s gross and operating margins are now exceptionally high by historical precedent; in fact, I would suggest that we’ve never seen a company generate a 60%+ operating margin at this scale of revenue and profitability. Again, it appears likely that defending these margins could be tough and we’ve seen peak monopoly power in the supply of GPU chips. Up and comers such as AMD look promising.

Nvidia operating margins: as good as it gets?

To be clear: we think that Nvidia still has a bright future ahead of it and can still deliver good returns to shareholders. However, this does not mean that it is THE market leader – that baton has likely been passed on to other players in the AI value chain. 

To cite a few examples of stocks that have worked exceptionally well recently: 

Palantir Technologies (PLTR)

Palantir provides advanced data analytics and AI-driven software platforms to large organizations, primarily in the government and commercial sectors. In the government sector, Palantir serves defense, intelligence, and law enforcement agencies (e.g., U.S. Department of Defense, CIA, FBI), providing tools for counterterrorism, cybersecurity, and mission planning, while in the commercial sector it assists companies to optimize supply chains, detect fraud, or improve operational efficiency. The company’s software platforms are especially adept at integrating disparate data sources and using AI to provide predictive analytics over a variety of use cases. 

Palantir’s stock gained 360% in 2024 and YTD has risen by a further 55% despite a major intra-quarter drawdown. Its valuation metrics, by all accounts, appear absurdly expensive at face value: the stock trades at a forecast EV / sales ratio of 76x, with the forward P/E multiple north of 220x. This is often what major market leaders look like though – crazy high valuations, which subsequently become justified by multi-year earnings delivery that surprises everyone to the upside. Of course, we can’t know for sure ahead of time that Palantir will still deliver to justify today’s price – but it certainly has the hallmarks of a market leader. To be sure, expectations are sky-high and any disappointments will be heavily punished along the way. Importantly, top-line growth is still accelerating - year on year, turnover has grown 21%, 27%, 30%, 36% and 39% respectively in the past five quarters. 

This is now a $300bn company which, in our view, has the potential to become the next “Microsoft” of the AI enterprise software industry. One of its limiting factors historically has been the slow sales cycle of its (often) multi-million dollar software contracts, but there is evidence that this is changing: in Q4 2024, Palantir reported a 71% year-over-year increase in U.S. commercial revenue (now 29% of total revenue vs 20% two years ago; this is no longer just a defense contractor play!) and a 65% rise in U.S. commercial customer count. This rapid growth indicates faster customer acquisition and deployment, suggesting a more efficient sales process. A likely contributor to this success is Palantir's Artificial Intelligence Platform (AIP) Boot Camps, which allow prospective customers to test and develop customized AI solutions in as little as five days; these were launched in mid-2023, and we can see the inflection higher in the rate of US commercial customer acquisition from later that year.

Palantir’s US commercial customer count – inflecting higher from Q4-23 onward

Palantir share price

Duolingo (DUOL)

Many readers will be familiar with Duolingo’s playful characters if they’ve tried to learn Spanish or German recently. The app’s trademark is the use of a gamified experience to make the process of learning languages more interactive. While there have been some concerns that AI would upend their business model (“why would I pay to use Duolingo if I can just ask ChatGPT how to ask for the menu in French?”), the evidence thus far suggests that AI is more likely a business enabler and accelerant for them. As a reminder, Duolingo’s offering includes a free-to-use basic version (which is ad-supported), with Super Duolingo offering an ad-free service for a subscription fee and access to more features. Duolingo’s newest top tier offering, Duolingo Max, offers AI-enabled conversational interaction with “Lily” which enables users to practice their speaking skills. This tier now accounts for 7% of their paid subscribers, and likely around 10-15% of bookings and revenue (it is a more expensive offering at $30/month in the USA). Duolingo cites significant gains in efficiency in publishing language course content due to embedding AI into their course creation processes – with the result that the rate of publication has exploded:

Source: Duolingo

It seems to be working: while Duolingo’s growth in users and subscribers has been strong for some time, Q1 2025 saw continued very strong growth in users and subscribers off an ever-rising base, with this quarter marking the single highest growth in daily active users in the company’s history (+6.1m users). An increasing proportion of users are converting to paid subscribers too – 8.9% of monthly active users (last 12 months) in Q1 2025, up from 6.8% three years ago. 

Source: Duolingo, author calculations

This is translating into real fundamentals: revenue growth of 38% year on year, with free cash flows up 31% year on year to $103m in Q1-2025. 

Sceptics will argue that AI poses a threat to Duolingo, the company appears to be doing very well at incorporating AI into their content design process and packaging and delivering their courses a curated way that still finds resonance with users such that they are still willing to pay. Furthermore, it is likely that Duolingo has potential way beyond just languages – they have already introduced maths, music and more recently, chess lessons. If they continue to execute well, Duolingo has the potential to become the go-to mobile educational platform for the AI era – beyond just languages, so the addressable market continues to widen. If so, at just 10.3m paid subscribers, this business could have a very long runway ahead of it. Again – much like other market leaders – you must pay for this potential, with the stock now trading at a forward P/E multiple of >80x and sporting a $22bn market capitalization.

After a 21% gain immediately post results, the share price has now broken out to new all-time highs. Not the sort of action from individual stocks that you tend to see in outright bear markets!

Duolingo share price

Apple: has the magic been lost?

Significant attention has been focused on Apple’s tenuous position in the face of Trump’s tariff assault on China. What was once lauded as Tim Cook’s crowning achievement – the development and refinement of Apple’s Chinese supply chain in producing its amazing products at scale for the global market – became a feared liability as far as its sales into the US are concerned. In its just-reported results, Apple highlighted to investors that it expects tariffs to cost them $900m in their June quarter – and that includes the benefit of sending all their Indian-made iPhone production to satisfy the majority of US iPhone demand. At the time of writing, this issue could potentially become a non-event given fresh news of a potential US/China trade deal and much reduced tariffs post meetings between the US and Chinese trade representatives in Switzerland this past weekend.

However, we think the bigger fundamental issue is the absolute lack of momentum in their core business of selling high-end devices. The iPhone is clearly key to this, but it extends further than this – their other devices have fared even worse in recent years, with sales levels below where they were four years ago.

Apple revenue: Services revenue the only driver of growth

While this trend has been great news for Apple’s gross margin percentage (due to mix effects; App store revenue attracts a much higher gross margin than device sales) and overall quality of earnings (more annuity revenue on an ever-growing installed base of devices despite slow device sales), it would be a stretch to call Apple a software business yet – Apple’s Services division today accounts for just over 40% of group gross profit dollars. This has been a material change in recent years (see chart below), but we would start calling this a software business when this metric materially exceeds 50%. What was missing from Apple’s recent results was evidence that the core engine driving the rise in the installed base of devices – new device sales (rather than simply devices ageing and being kept longer!) – was meaningfully growing again. Instead, Apple continued to deliver pedestrian Product revenue growth of 2.7% year on year, with iPhone <2%. 

Apple: % of gross profit attributable to Services division

Source: Apple disclosures, author calculations

The trouble is that Apple’s products – iPhone in the main – appear to have lost some of the “magic” that their very loyal customers typically associate with the upgrade process. The new phones just don’t seem that much better than the ones customers would already have, and the changes have been far more evolutionary than revolutionary. At the same time, there is a high chance their device business has become a victim of its own success: given advances in battery technology and life, the existing cohort of iPhones in people’s pockets last so long that the need to upgrade just isn’t there like it may have been 5-10 years ago. 

Luckily, Apple’s installed base of iPhones in use today continues to grow and likely exceeds 1.5 billion devices globally (and this is great for growing their higher-margin Services revenue stream) – but this has been more about lengthening replacement cycles than a boom in annual sales (i.e. proportionately more older phones remain in use, while unit sales stagnate). 

Apple doesn’t disclose unit sales anymore, so as investors we are forced to make educated guesses – but based on our research we would guess the average age the iPhone installed base today exceeds 4 years, compared to probably less than 3 years as recently as five years ago. Many devices in use are much older. This could be a taken as a bullish sign given the “elastic band” theory – the longer you stretch an elastic band, the more violently it snaps back. The more the installed base of iPhones ages, the closer we are getting to a monster replacement cycle of old devices. We just haven’t seen evidence of this yet, and investor patience is running thin given how expensive Apple’s shares are – a forward 30x P/E, with sales growing just 5% per annum. Excellent capital management takes you so far – but eventually, organic growth in the core profit engine needs to return.

As a result, Apple is no longer a market-leading stock – it has delivered returns broadly in line with the S&P500 for almost five years now:

Apple share price relative to S&P500 10-year history: a tale of two halves

We think Apple is a great quality company that continues to deserve a place in portfolios; we are just not convinced that it is likely to deliver market-leading returns from here.

Alphabet – facing existential threats for the first time in its existence

The market is very worried about Alphabet – this much is obvious from the steady collapse of its P/E multiple relative to the S&P500 in recent months, as well as a 7% drop in its share price earlier this week on testimony by Apple’s SVP for Services that it had seen a decline in search volumes on its browser for the first time ever in April.

Alphabet forward P/E relative to S&P500:

This is now a company that the stock market is, rightly or wrongly, treating as a “melting ice cube.” 

The issue?

Google has enjoyed a near-monopoly on internet search for 20+ years, and for the first time is facing the risk of its auction-driven search result business model being disrupted by AI-newcomers entering the search arena. To be sure, they have their own versions of AI apps, such as Gemini – but that misses the point: they have little to gain here in relative terms and are in a position of having to defend a large existing profit pool (global search ad revenues), rather than the upside of starting from scratch in creating a new one. I’ve noticed my own habits changing – I use Google search far less than I used to, deferring to Grok as my go-to app, especially when conducting research (it is very useful for summarising earnings call transcripts!). It was previously unthinkable that habits like this could be broken, but it appears to be happening. 

Alphabet does have other drivers of value these days – their Cloud business is growing rapidly, and while their Other Bets division remains loss-making, it contains some promising assets that could be very valuable (e.g. Waymo). The issue is one of relative contribution: Cloud profits remain only 6% the size of Google Services, while the latter’s profits are still dominated by Search. As far as profits today are concerned, Alphabet is still a one-trick pony.

And that pony has showed up to the horse show with competitors showing off some sexy looking young new foals for the first time ever.

To be sure, at this rating, Alphabet can still deliver decent returns in a scenario of declining share of internet advertising dollars provided the market continues to grow robustly. But until they can demonstrate definitively that the existential threat concerns are misplaced (via stabilising or even growing market share again), it is likely the stock’s prospective returns are unlikely to see much help from improvements to its rating. In short: it is unlikely to be a market leader.

Conclusion

In 2024, it was all about the Magnificent Seven stocks – if you didn’t own them in healthy size, nothing else that you did mattered much and it’s likely you underperformed broader indices. 2025 has brought a not-so-subtle shift – we’ve seen sharp underperformance of this cohort of stocks YTD, but we don’t think this is because the overall AI theme is over. Rather, the issues that have driven underperformance for the “Mag 7” are likely idiosyncratic, company-specific issues (some of which may well be resolved in time), while we’ve demonstrated that there are examples of strong “tomorrow’s winners” in AI coming through the ranks. That’s how markets work – the leadership baton gets passed on. 

Being more selective in 2025 is likely to pay off.

Magnificent Seven ETF (MAGS) price relative to S&P500 (SPY ETF)

Related Stories

No stories found.
BizNews
www.biznews.com