The ‘Magnificent Seven’ stocks spent years convincing investors that their tremendous AI spending would translate into sustainable growth, powerful cash flows, and fatter valuations.
That assumption is now up against a major test.
Wall Street has largely treated hyperscaler spending as a powerful long-term growth engine, but Bank of America strategist Michael Hartnett just flagged a major risk that could test how much investors are willing to pay for the AI trade.
The contrast is becoming incredibly tough to ignore.
Stock markets remain somewhat resilient, but parts of the credit market are flashing more caution around AI spending.
That said, BofA now sees one major market signal as critical to the Mag 7’s ability to shrug off that threat.
Why BofA sees a critical test for the Magnificent Seven
According to Seeking Alpha reporting, Hartnett just identified what needs to keep working for the AI trade to remain credible.
More AI:
- Nvidia just made a move Wall Street wasn’t ready for
- Microsoft just took sides in AI policy fight
- OpenAI just disclosed something genuinely alarming
He zeroes-in on the Roundhill Magnificent Seven ETF (MAGS), holding around $70, turning the ETF into a confidence gauge.
If the group can maintain pricing strength despite concerns about cheaper Chinese compute, it suggests investors are still buying into the long-term AI CapEx story.
For now, though, it seems markets are becoming a lot less comfortable with that assumption.
Credit spreads and credit-default swaps linked to AI hyperscalers are moving in a far more cautious direction.
He pointed to two major signs, including rising U.S. investment-grade tech credit spreads and Oracle’s (ORCL) five-year CDS, as evidence that credit investors are growing increasingly cautious about the AI infrastructure trade.
Though stocks are still rewarding the AI story, credit markets are beginning to question its cost.
Cheap Chinese computers add a major dynamic to that layer.
If increasingly capable AI can be developed and operated at significantly lower costs, U.S. hyperscalers will need to justify why hundreds of billions of dollars in annual capital spending is necessary.
For investors, the biggest risk might therefore be valuation compression instead of an immediate earnings collapse.
If we see confidence in the CapEx cycle weaken, investors might demand lower multiples before sales or earnings materially deteriorate.
The key signals to watch are MAGS price strength, hyperscaler cash flow and buybacks, and credit spreads.
If stocks are depressed while credit stress continues rising, the market could be starting to question the economics behind the AI boom, not merely its near-term growth rate.
Why is cheap Chinese compute a threat to the AI capex boom?
Chinese AI developers are showing they can deliver highly capable models using cheaper hardware, more efficient architectures, and dramatically lower inference costs.
The issue is that it runs counter to one of the assumptions underpinning the U.S. AI boom, which entails that better AI will require ever-larger amounts of expensive computing infrastructure.
DeepSeek first exposed that flaw in early 2025.
CNBC reported that its V3 model was developed using less-advanced Nvidia H800 chips, citing training costs of under $6 million.
The reaction was immediate, with investors questioning whether U.S. companies really needed to shell out billions in building the AI ecosystem.
Consequently, according to CNBC, Nvidia dropped nearly 17% on Jan. 27, 2025, wiping $593 billion from its market value in a single session.
That threat has only gotten more tangible over time.
DeepSeek’s new V4-Flash costs just $0.14 per million input tokens and $0.28 per million output tokens, according to Artificial Analysis data reported by Reuters.
Even though it was remarkably cheap, the model was much more competitive than more expensive systems, including Alibaba, Z.ai, Moonshot, and ByteDance.
So if businesses can achieve similar AI performance with far fewer GPUs or cheaper models, the economic return on those billions of dollars becomes much less certain.
There’s clearly a ton of financial pressure involved already.
Big Tech’s AI buildout is now up against a major cash-flow problem, while Microsoft, Meta, Oracle, Amazon, and Alphabet have collectively written down nearly $1.09 trillion in future lease payments, much of it tied to data centers, according to Reuters.
Moreover, credit markets are noticeably more cautious, with Oracle’s five-year credit-default swaps trading around 200 basis points, compared with nearly 53 basis points for a broader investment-grade CDS index, according to Reuters.
There’s the counterargument, too, though, which entails that cheaper AI could raise compute demand.
Yahoo Finance reports that Microsoft CEO Satya Nadella made that argument following the original DeepSeek shock, arguing that greater AI efficiency will drive significantly more demand.
So far, we haven’t seen the U.S. hyperscalers respond by slashing spending.
Case in point: Amazon recently raised its 2026 capex forecast to $220 billion, citing healthy AWS demand and ongoing capacity constraints, according to the Financial Times.
What are the Magnificent Seven stocks?
The “Magnificent Seven” refers to seven of the most powerful U.S.-listed companies that have become the most prolific stocks in the S&P 500.
Their influence is down to their sheer size, top- and bottom-line strength, and exposure to trends such as AI, cloud computing, and digital advertising.
According to Reuters, the term was coined by Hartnett in May 2023, a nod to the 1960 Western The Magnificent Seven.
These seven stocks have created tremendous wealth in the stock market.
According to NYU finance professor Aswath Damodaran, their combined market capitalization jumped by $5.1 trillion in 2023 alone. That accounted for over 50% of the increase in the value of the entire U.S. stock market that year.
Moreover, a separate FTSE Russell report showed their combined market value jumping another 43.5%, from $9.2 trillion to $13.2 trillion, over the year through April 30, 2024.
The Magnificent Seven are:
- Nvidia (NVDA) — $5.424 trillion.
- Apple (AAPL) — $4.572 trillion.
- Alphabet (GOOG) — $4.322 trillion.
- Microsoft (MSFT) — $3.712 trillion.
- Amazon (AMZN) — $2.960 trillion.
- Meta Platforms (META) — $1.508 trillion.
- Tesla (TSLA) — $1.297 trillion.
Source for market caps: CompaniesMarketCap, as of Aug. 7, 2026.
How does MAGS’ performance stack up against its risk?
The Roundhill Magnificent Seven ETF has slowed significantly compared to its lofty year-over-year gains, according to Seeking Alpha.
The ETF has gained 4.82% over the past week and 4.38% over one month, outperforming the S&P 500’s 3.57% and 3.38%, respectively.
However, things look a lot shakier over a six-month period, with the ETF posting a 9.55% gain compared to the broader market’s 12%.
Longer-term, however, Roundhill has been a massive money spinner, surging 121% over three years compared to 71.7% for the S&P 500, according to Seeking Alpha data.
It’s important to note, though, that the MAGS carries a higher-than-average risk profile, with its holdings heavily concentrated in a small group of stocks.
Nearly 96% of its assets sit in its top 10 holdings, double the typical ETF level of 45%. Its annualized volatility of 22.2% is also well above the ETF median of 14.1%, underscoring larger price swings.
Moreover, its standard deviation is elevated at 24, compared to the ETF median of 13, reinforcing that MAGS can move much more sharply than the average ETF.
Related: Warren Buffett keeps pointing at the same ETF for a reason
