The Great Divergence: Q2 earnings and the unravelling of the ‘Magnificent 7’

The current phase of earnings reporting has marked a shift in the paradigm. Confidence in the productivity growth promised by AI has started to falter. Investors have started to demand something the market spent two years deferring: profit. They have realised that the unprecedented Capex on AI by companies needs to be paired with a semblance of ROI otherwise the promised future earnings can no longer be justified. This leaves the MAG 7, particularly the hyperscalers, in a precarious balancing act where too much capex, without the accompanying growth, leads to a loss in confidence, whereas not enough capex risks obsolescence, with the rhetoric ‘AI or Die’ becoming ever more prevalent. So far, we have seen 6 out of the 7 ‘Magnificent 7’ (MAG 7) companies (Alphabet: GOOGL, Tesla: TSLA, Microsoft: MSFT, Meta: META, Apple: APPL, Amazon: AMZN) report their Q2 (April, May, June) earnings and in this article I will try to make sense of why the market reacted to each report the way it did.

Alphabet: GOOGL

Alphabet released earnings on July 22^nd. The top line was strong, showing total revenues hitting $119.8bn (24% growth year-on-year) which beat expectations of $116.9bn. Notably Google Cloud revenues surged 82% to $24.8bn which smashed the street’s estimates of 63%. The unprecedented growth was the result of an unrivalled full stack position. Alphabet owns the infrastructure layer through its custom TPU silicon, the model layer through Gemini and the application layer through Search and YouTube. The company leans heavily on this vertical integration noting that the TPU 8i and 8t chip designs are central to the infrastructure buildout whilst also claiming that TPU system sales are part of the cloud sales backlog which reached $514bn in Q2 growing more than $50bn since Q1 and revenue from those systems are expected to ramp up primarily in 2027.

However, the buildout has been enormous. Alphabet disclosed $44.9bn in capital expenditure for Q2 which is double Q2 last year and up $35.7bn from the previous quarter. On top of this, CFO Anat Ashkenazi raised the full year 2026 capex guidance to $195bn-$205bn, up from the $180bn-$190bn range set after Q1 and cautioned that 2027 capex spending would ‘significantly increase’ further. The company has already deployed $80.6bn in capex through the first half of 2026 implying a run rate of $60bn over the next 2 quarters, suggesting spending is accelerating not plateauing. Ashkenazi broke down the spending into a 60/40 split: 60% (around $27bn) went to servers, mainly AI accelerators like Google's proprietary TPU chips and Nvidia GPUs: 40% (around $18bn) funded data centre construction and networking equipment.

Unfortunately for Alphabet, the lack of fiscal discipline has had a detrimental effect. For the first time since Alphabet went public in 2004, free cash flow turned negative with the $44.9bn capex exceeding operating cash flow of $39.1bn resulting in negative $5.9bn free cash flow. On top of this, management acknowledged pressure on the profit margin in Google Cloud during Q3 due to the use of third-party capacity as a bridge while owned infrastructure comes online and, although the growth was impressive, Google Cloud’s operating income accounted for only 21.6% of total operating income. These issues overshadowed the positives and led to shares falling about 4% in after-hours trading.

Tesla: TSLA

On the same day as Alphabet, the Q2 2026 earnings report from Tesla was released. Record overall revenue of $28.2bn solidly beat analyst’s estimates of $26.7bn and resulted in Tesla crossing $100bn in trailing twelve-month revenues for the first time. This was driven by a record quarterly vehicle delivery of 480,126 units (up 25% year-on-year) with automative revenue rising 23% to $20.5bn, whilst the Services and Other segment (consisting of all non-core automotive manufacturing and energy storage sales) jumped 50% to $4.6bn.

However, this was overshadowed by operating income falling by 57% to $398mn, squeezing profit margin from 4.1% a year earlier to 1.4%. What is striking is that at Tesla’s peak in 2022 margins were about 17%, exemplifying the struggles the company has been facing over recent years. Part of this erosion has come from competitive pressure from BYD which delivered 557,090 battery electric vehicles in Q2, exceeding Tesla by around 77,000 units, with BYD outselling Tesla in Europe for several consecutive months. But much of the erosion can be attributed to operating expenses rising 47% to $4.4bn as the company increased spending on AI infrastructure, the robotaxi program, and Optimus robot production.

On top of this, Tesla expects its full-year 2026 capital expenditures to exceed $25bn, an increase from earlier estimates of $20bn, with Musk justifying the $5.9bn spend in Q2 as necessary to become the market leader in robotics and automation. Musk is asking investors to look past the car company entirely and value tangible milestones such as robotaxis and Optimus whilst also hinting at a potential merger with SpaceX.

Free cash flow swung to negative $1.09 billion, compared with positive $146 million a year earlier. Tesla has not generated positive free cash flow in any quarter of 2026.

Tesla’s lacklustre operating margin and Musk’s empty promises did not fill investors with confidence and led to a drop in share price by 6% in after-market trading with a collapse of 18% by the end of the week. Tesla (TSLA) became the biggest loser in the S&P 500 on July 23, 2026, plunging nearly 15% and wiping out $214.5bn in market value in its worst single day drop on record. Some people made the right bet, and some short sellers were poised to make $4bn in one day on a mark-to-market basis.

Microsoft: MSFT

A week later on July 29^th, Microsoft reported its Fiscal Year (FY) 2026 Q4 earnings. A revenue of $90bn, indicating 18% growth year-on-year, smashed through expectations by $2.4bn. Adjusted earnings per share of $4.74 topped the $4.24 estimate, though analysts noted that roughly 33 cents of the 50-cent beat came from a $3.2bn gain on Microsoft's Anthropic investment stake.

Azure (Microsoft’s cloud segment) was the headline, with growth accelerating to 43%, up from 40% in the previous quarter and beating expectations of 40% whilst maintaining an operating margin of 45%. This was the fastest rate of growth in 4 years and was accompanied by a milestone of Azure crossing $100bn annual revenue for the first time. This was driven through the different segments of Azure, with Microsoft 365 Copilot reaching 30mn paid seats, up from 20mn disclosed in April, representing an estimated $10bn annual run rate. The outlook looked promising with Azure guided to accelerate further to 45% growth next quarter supported by commercial remaining performance obligations (a measure of future contracted revenue) jumping 84% to $678bn, signalling sustained enterprise demand for computing infrastructure and AI services.

The strong performance of its segments was supported by prudent fiscal discipline. A positive free cash flow of $19.6bn, although a 23% year-over-year decline, represented a commitment to investors and was supported by Q2 capex coming in at $41bn, below the $42bn feared by analysts. Also, the company guided 2026 capex down to $175bn from $190bn. The reduction came from financial jiggery pokery rather than an actual reduction in capex as Microsoft extended the estimated useful life of its data centres and office buildings from 15 years to 25 years, effective at the start of fiscal 2027. That longer useful-life assumption changes how certain contracts get classified. More future data-centre leases get classified as operating leases (instead of finance leases) which do not count toward capex.

In after-hours trading, the stock rose 3% and on July 30^th surged 15% which was the sharpest single-day performance gap between two mega cap tech stocks in recent memory, with a combined swing of roughly 23 percentage points between Meta and Microsoft on the same trading day.

Meta: META

Alongside Microsoft we had Meta release its Q2 earnings report. From the get-go, Meta has a structural weakness from its lack of a cloud business which means everything comes from its advertising business. Fortunately, Meta’s top line cleared the bar with $60.8bn in Q2 revenue exceeding the $60.2bn consensus. This was powered by a 27% increase in advertising revenue to $59.4bn with ad impressions rising 14% and average price per ad climbing 12%.

That was the end of the good news for Meta. Adjusted earnings per share came in at $6.18, missing the $7.17 Wall Street consensus by 14%. Operating margin narrowed from 43% to 31% as total costs and expenses surged 55% year-on-year to $42bn. Net income declined 14% despite the revenue growth. The most alarming figure was free cash flow which was at its lowest in years after ballooning expenses have been eating into AI bets. Meta's free cash flow collapsed 91% from $8.55bn a year prior, to $784mn as Q2 capex of $31.1bn consumed 97.5% of operating cash flow. First half capex totalled $50.9bn, and the company raised its full year capex guidance to $130bn-$145bn, up from the previous lower end of $125bn, requiring average quarterly spending roughly 27% above Q2 levels in the second half. CEO Zuckerberg hinted at a potential cloud business to monetize surplus compute, but CFO Susan Li declined to provide specifics on 2027 cost trajectories, noting only that Meta expects to remain 'demand constrained' for computing power.

The stock, which had already been falling for nine consecutive sessions, dropped another 9% on July 30, bringing the cumulative 10-day decline to over 14% while Microsoft posted its best single-day gain since March 2020. Q3 revenue guidance compounded the damage as Meta guided to a midpoint of $62.5bn, below the $63.2bn analyst consensus. Investors will be looking out for efforts to build cloud business to open new revenue streams and offset aggressive capex.

Apple: APPL

The next day on Thursday 30^th July we had Apple report its earnings for FY Q3 2026. Tim Cook’s last earning report was disappointing with the only good news being overall revenue of $109.4bn beating Wall Street expectations of $108.7bn. Despite the 22% growth in overall revenue, two key business areas sparked concern. Sales in China grew slower than expected with revenue coming in at $18.8bn compared to a $19.6bn expectation. On top of this, Apple Services reported revenues of $30.7bn, below estimates of $31.2bn with growth in this segment expected to decelerate in Q3 due to the impact from regulatory changes in the app store model in the EU. Weakness in the services area is particularly worrying since the subscription-based cash flows account for so much of Apple’s premium PE (price to earnings) ratio.

The main theme was uncertainty remaining about the memory chip shortage and how apple was navigating this aside from higher prices. Apple is battling companies like Amazon for the components as AI and the iPhone maker are both trying to get as much hardware as they can. Tim Cook stated that the supply shortage would affect MacBook, iPad and iPhone this quarter. The supply crunch has left longer waiting times for computers and more importantly the revenue growth forecast was little bit weaker than expected. Fiscal Q3 earnings were okay but looking ahead apple said revenue will rise 9-11% which is lower than estimates of 12%, which stems from the supply constraint.

Even though free cash flow was $32bn which is more than Microsoft, Meta, Alphabet and Amazon combined, this was not rewarded by investors. Stock price fell 4.7% in the aftermarket and the next day Apple suffered its worst stock decline since March 2020 with market cap falling by $500bn and resulting in Apple being the worst performing in the Dow industrials. However, it is not all doom and gloom, Apple has been the best performing out of the Mag 7 this year (being up more than 20% this year) and following the drop the share price is still up 11% so far this year.

Amazon: AMZN

The highlight of the earnings quarter so far has been Amazon. It released earnings alongside Apple and delivered what the market had been looking for. The good news started from the top line where total revenue increased by 20% to $200.6bn beating expectations of $197bn. This was followed by a 13.7% operating margin also beating estimates of 12%.

The star of the show was its cloud service: AWS (Amazon Web Services). Posting accelerating cloud revenue for the 5^th consecutive quarter helped ease concerns about AI bets not paying off. Revenue jumped 37% year-on-year to $42.2bn smashing through the expected $40.6bn and posting the highest rate of growth since Q4 2021. Operating income surged to $16.6bn up from $10.2bn in Q2 2025 with an operating margin of 39.4% and accounted for 60.5% of operating profit for Amazon meaning AWS is the leading source of profit growth. AWS was one of the first to develop cloud service for enterprise meaning they have a more diverse customer base and have been doing it for longer than its competitors which helps explain part of the standout profitability. CEO Andy Jassy highlighted the stellar growth and that this was driven partly by the AI and chip segment of AWS which achieved a milestone of $25bn run rates. This is all supported by a backlog in demand of $496bn Q2 2026 which rose from $364bn in Q1 2026 and marked an increase of over 150% year-on-year.

There were a few sore spots that investors chose to ignore. Physical store revenues were below estimates of $5.9bn, coming in at $5.8bn. This was due to concepts like Amazon Fresh and Amazon Go failing to attract a loyal customer base. Out of all the capex stories, Amazon had the worst one. Having spent $53bn on property, plant and equipment meant Amazon reported the lowest free cash flow out of all companies who reported so far, with an outflow of $7.6bn. This was accompanied by an upward revision in capex projection for 2026 from $200bn to $220bn with CEO Andy saying most of that spending will go to AI and the increase was driven mostly by the upward pressure from component costs particularly memory chips. Q3 forward guidance also disappointed with estimates between $197bn and $202bn compared to the $203.9bn expected. However, the Q2 results and future investment plans eased concerns about how much Amazon is spending to meet increasing demand for AI. Part of this spending was a 1bn investment for AWS forward deployed engineering with the concept being AI engineers embedded directly with customers to co-develop and deploy AI agentic solution in days rather than months. This has already been supported by demand with early customers including Allen Institute, NBA, NFL, Cox Automotive, Ricoh and Southwest airline. On top of this spending Amazon sold $62bn in bonds this year solidifying the commitment to building out AI.

The market reaction to the report was stratospheric. The stock went up 8% in aftermarket trading and continued to rise. By the end of the week shares were up 15% adding $389bn in market cap and marking its best performance since April 2012 and claiming the biggest gainer position in S&P 500 and Nasdaq 100. Then, on Monday, it reached a historic milestone only NVIDIA, Alphabet, Microsoft and Apple have achieved: a $3bn market capitalisation.

What is interesting to note is that Alphabet reported significantly higher growth in its cloud service alongside lower spending and higher free cash flow but did not receive the same reaction. This could be since most operating profit comes from AWS (60.53%) whereas Google Cloud only makes up 21.61% of total operating profit. Morgan Stanley wrote AWS delivered accelerating growth, surging profitability and a better backlog giving more confidence for the market on return on investment for the cost ahead.

Overview

To conclude, the companies that were rewarded, Microsoft and Amazon, have created a second engine in the form of AI cloud services that are able to turn capex into a visible return. The companies that were punished, Tesla and Meta, asked the market to fund enormous spending on the promise of returns that have not yet materialised. And Apple, ended up being punished for a shortage the race created.