Key points
- Bills passed by the US Congress and trade agreements lifted global shares though added to future US debt levels
- The Q2 earnings season was broadly positive, supporting the outperformance of large-cap tech stocks
- AI-led investment accelerated and increasing AI adoption pushed greater levels of spending, particularly on data centres
Market review
Equity markets continued to rally in July, with returns augmented by strength in the trade-weighted1 US dollar index which gained +3.2%. During the month, the MSCI All Country World Net Total Return Index rose 5.0%, in sterling terms, while the NASDAQ made 14 record highs during the month as the US signed trade deals with key partners and tech earnings reports were supportive.
On 4 July, President Trump’s One Big Beautiful Bill Act passed the Senate, providing a stimulus package that could offset some of the potential drag from tariffs. Bitcoin rose +10.1% during the month on the passing of the GENIUS Act, which provided a regulatory framework for stablecoins2, while Brent crude oil was up 7.3% on a more positive global growth outlook and potential supply concerns given conflicts in the Middle East and Europe.
The ‘Big Beautiful Bill’ marked a significant stimulus package for US corporates and should boost free cashflow through tax cuts and incentives for business investment. The Bill is expected to add $3-4trn to the US deficit over the next decade but should support corporate capital expenditure into 2026. Having extended the tariffs deadline to 1 August, the US signed key trade deals with Japan, South Korea and the European Union with rates set at 15% in return for commitments to US investment of several hundred billion dollars.
US economic data remained strong with 147,000 jobs added in June. However, the July jobs data released on 1 August suggested a meaningfully softer labour market, albeit still with historically low unemployment. The longer-term impact of tariffs is uncertain, but June CPI data showed some price increases coming through despite core CPI coming in lower than expected. The Federal Reserve’s (Fed) preferred inflation metric, core Personal Consumption Expenditures (PCE), which excludes volatile food and energy prices, increased +0.3% month on month, up from +0.2%, but the overall inflation backdrop remains relatively benign.
Fed Chair Jerome Powell clashed with the US president during the month, stoking fears once again that Trump would try to fire him. The FOMC left the target rate unchanged at 4.25-4.5%, as widely expected, but Governors Bowman and Waller dissented in favour of a 25 basis point3 (bp) cut. As potential successors to Powell, this move was seen as a nod to Trump’s pressure on the Fed to cut rates and potentially an early move to succeed Powell. This marks the fifth consecutive pause in the Fed’s current easing cycle, even as the central bank acknowledges that the unemployment rate remains low and labour market conditions are still solid.
August started on a challenging note, with equities falling sharply and US dollar and Treasury yields sliding on considerably weaker than expected July employment data, with only 73,000 jobs added (below expectations for 110,000) and June data revised sharply downwards to 14,000 (rather than 147,000 previously reported). Interest rate expectations shifted markedly, with an 87.5% probability of a September cut priced in versus 37.7% the previous day. Market weakness was likely exacerbated by fresh tariff announcements on 1 August as well as Trump firing the head of the US Bureau of Labour Statistics, stoking concerns over Fed independence and US policy when Powell steps down in 2026.
Technology review
The technology sector continued to outperform the broader market in July, with the Dow Jones Global Technology Net Total Return Index (W1TECN) returning +8.0%, in sterling terms.
Large-cap technology stocks materially outperformed their small and mid-cap peers. The Russell 1000 Technology Index (large cap) and Russell 2000 Technology Index (small cap) returned +5.5% and -0.1% respectively. The Philadelphia Semiconductor Index (SOX) returned +1.2% while the NASDAQ Internet Index (QNET) and iShares Software (IGV) returned +0.9% and +2.1% respectively (all returns in USD terms).
Fundamental AI model progress continues at a rapid pace as xAI released Grok 4 which Elon Musk claimed was “the smartest AI in the world”, the model achieving a record score of 25.4% on Humanity’s Last Exam (HLE), an AI benchmark, and 44.4% in ‘Grok 4 Heavy’ mode which uses multiple agents. OpenAI launched a ChatGPT agent which combines deep reasoning (Deep Research) and autonomous web browsing (Operator) to deliver agentic capabilities. It scored 41.6% on HLE, a significant improvement on Deep Research’s 26.6%.
ChatGPT was the most downloaded app in the world for the second month (50 million downloads) and reportedly reached 700 million weekly active users (WAU) in July
AI adoption also remains strong. ChatGPT was the most downloaded app in the world for the second month (50 million downloads) and reportedly reached 700 million weekly active users (WAU) in July, up from 500 million in March. The company apparently also raised $10bn in June and another $8.3bn ahead of schedule in July and is operating at a $12bn annualised recurring revenue4 (ARR) run rate. Rival Anthropic has reportedly reached $4bn in ARR and is in talks to raise another round of funding at a $170bn valuation, against OpenAI’s reported $300bn. Google DeepMind’s CEO, Demis Hassabis, reported that Alphabet processed nearly a thousand trillion (a quadrillion) tokens in June, more than double the amount processed in May, as Google rolls out its AI search experience and its reasoning models drive significantly higher tokens per use (Google began rolling out its Gemini 2.5 reasoning models at the end of Q1). MetaAI has more than one billion monthly active users, up from 700 million at the start of the year.
Earnings season was broadly positive – at the time of writing, 68% of S&P 500 companies have exceeded revenue forecasts while 81% beat on earnings. In the internet sector, Alphabet’s results met elevated expectations with Search up 12%, YouTube 13% and Cloud 32% year on year (y/y). Management highlighted AI’s broad impact, citing strong traction from new search features like AI Overviews (two billion monthly active users (MAUs), 10% query growth) and AI Mode (100 million users in US/India). Gemini reached 450 million MAUs while Veo 3 created 70 million videos. Cloud growth accelerated, with backlog up 38% and easing supply constraints expected in 2H25. Capital expenditure (capex) guidance for 2025 was unexpectedly raised by $10bn to $85bn, with further growth expected in 2026 to support strong Cloud and AI demand.
Meta Platforms (Meta) delivered a strong quarter, with revenue growth accelerating to +22% y/y (6% above consensus), noting that “AI-powered recommendations improved ad conversions by c5% on Instagram and c3% on Facebook”. Next quarter guidance for 17-24% y/y revenue growth was also well above expectations. CEO Mark Zuckerberg said Meta is aiming to build self-improving AI, with superintelligence “now in sight.” The company continues to invest heavily in the AI opportunity with 2026 capex expected to increase by a further $30bn from $69bn this year.
Microsoft also posted strong results as Azure revenue growth accelerated to +39% y/y, well above guidance of+34-35% y/y, and guided to +37% y/y growth next quarter. While management did not break out the AI contribution to growth, and will no longer do so going forward, demand remains strong beyond just OpenAI, leaving Azure capacity constrained through calendar year-end. Copilot tools have reached 100 million MAUs (including 20 million for GitHub Copilot) and 800 million monthly users have used some AI functionality. Capex is expected to be $30bn next quarter, $6bn ahead of consensus estimates, to help fulfil the company’s remarkable backlog which grew +35% y/y to $368bn.
Amazon posted mixed results, with strength in retail (+12% y/y) and advertising (+22% y/y) offset by lacklustre growth at Amazon Web Services (AWS; +17.5% y/y) – disappointing considering strong performance at peers Google Cloud Platform (GCP) and Microsoft Azure. In addition, the company did not commit to a re-acceleration of AWS this year, fuelling concerns about its AI positioning. Management noted that it is still “very early” and is hopeful that it will regain incremental market share as AI spending transitions from training to inference. Amazon’s capex surged to $32bn in the quarter (well above forecasts of $26bn) and management raised full-year capex guidance from $105bn to $115-120bn.
In semiconductors, TSMC posted strong results, with revenue up +39% y/y and margins ahead of expectations
Spotify Technology delivered mixed results. The company reported strong underlying demand trends, with eight million premium subscriber additions in the quarter but revenue only grew +10% y/y (hindered by currency headwinds) and operating profit missed expectations due to higher staff and marketing costs. Netflix posted a solid quarter and issued guidance above consensus expectations but this was outweighed by concerns about declining per-member engagement.
In semiconductors, TSMC posted strong results, with revenue up +39% y/y and margins ahead of expectations despite foreign exchange and overseas fab5 headwinds. AI demand remains robust, with high-performance computing up +60% y/y, while smartphones also surprised positively, likely due to tariff pull-ins. Q3 revenue guidance of +8% quarter-on-quarter (q/q) exceeded expectations and FY25 growth was raised to +30% y/y.
NVIDIA and Advanced Micro Devices (AMD) did not report in the month but benefited from upwardly revised hyperscaler6 capex intentions as well as the Trump administration easing restrictions on the sale of AI semiconductors into China.
Several other AI-related positions performed well too. GE Vernova reported revenue and EBITDA7 well ahead of expectations while management raised full-year guidance, supported by the equipment backlog which grew +10% q/q to $50bn. In the electrification business, the company received $500m of orders from hyperscalers in the first half of the year, compared to $400m in the whole of 2024. Global electronics manufacturing services provider Celestica also reported strong results, with revenue up +21% y/y (compared to consensus estimates of +12% y/y) driven by its hardware Platform Solutions business which grew +82% y/y due to strong AI-related demand for networking components.
Apple underperformed during the month given ongoing concerns about the potential impact of tariffs as well as concerns around the company’s AI strategy. However, the company delivered stronger than expected results and guidance. Revenue was +10% y/y (6% above consensus) in the quarter, with key iPhone, Greater China and services areas all better than expected. There was a currency tailwind and management indicated a modest pull forward of demand due to tariffs. Commentary hinted at rising AI investment, as well as the potential for an AI companion device, although the long-term AI strategy remains unclear.
Outlook
The fastest adopters are seeing AI impact revenue and cost bases already. Meta is still seeing strong returns on its internal AI spending, as AI drove +5% more ad conversions on Instagram and +3% on Facebook, and AI-powered recommendation advancements led to a +6% increase in time spent on Instagram and +5% on Facebook this quarter. Microsoft saved more than $500m in call centre costs alone last year using AI while increasing customer and employee satisfaction, and 35% of code for new products is AI-generated. Axon Enterprise found that law enforcement customers reported saving between six and 12 hours per week using its AI tools.
NVIDIA, Meta and Microsoft leading the group as perceived AI winners while AI concerns have weighed on Alphabet and Apple
We were also encouraged by a recent Federal Reserve paper that supports our view of generative AI (GenAI) as a general-purpose technology (GPT). GPTs are a unique class of innovations characterised by widespread adoption, the ability to drive sustained complementary innovations and continual improvement – traits the paper attributes to GenAI. Moreover, the paper suggests GenAI may also qualify as an invention of methods of invention (IMI) – technologies that enhance the efficiency of research and development by improving observation, analysis, communication or organisation, much like the invention of the microscope did. Both GPTs and IMIs have historically been linked to broad-based productivity gains, which in turn can support structurally higher GDP (gross domestic product8) growth and lower inflation. According to the Congressional Budget Office, such productivity-driven growth could also play a critical role in addressing the long-term fiscal challenges facing the US.
While large-cap technology stocks have continued to meaningfully outpace small-caps, there has been a notable bifurcation of performance within the sector and the Magnificent Seven (Mag79) with NVIDIA, Meta and Microsoft leading the group as perceived AI winners while AI concerns have weighed on Alphabet and Apple. Recent weakness at Tesla reflects weaker near-term fundamentals as well as the high-profile deterioration of the relationship between CEO Elon Musk and President Trump.
As a reminder, the Trust remains firmly focused on the AI opportunity with significant exposure to AI enablers encompassing compute, memory, storage, networking, power and cooling. Rapid AI progress and continued confidence in AI scaling has supported significant increases in AI investment. OpenAI signed a 4.5 gigawatt (GW) data centre agreement with Oracle and expects to exceed the initial Stargate commitment of $500bn and 10GW over the next four years. For context, Microsoft stood up just over 2GW of data centre capacity globally during the past 12 months. Meta CEO Mark Zuckerberg expressed his plan to spend “hundreds of billions” of dollars to build Superintelligence across several multi-GW superclusters. Anthropic believes it will require a 2GW data centre in 2027 and 5GW in 2028 to train advanced models, and that total AI training power demand will reach 20-25GW by 2028.
AI investment on this vast scale may add up to $3trn in global data centre spending (not including related power investments) through 2028 and a 6x increase in capacity by 2035, according to Morgan Stanley. Furthermore, a talent war has broken out as Meta is on an AI hiring spree. Despite this investment response, AI demand is still outpacing supply, according to Microsoft, Alphabet and Amazon. Microsoft CFO Amy Hood was unequivocal: “In January I said I thought we’d be in better supply-demand shape by June. And now I’m saying, I hope I’m in better shape by December. And that’s not because we slowed capex – even with accelerating the spend and trying to pull leases in and get CPUs [central processing unit] and GPUs [graphics processing unit] in the system as quickly as we can, we are still seeing demand improve.”
Aggregate cloud revenue growth across the hyperscalers accelerated +3pts q/q to +28% and overall backlog grew +34%, according to Bank of America. In 2025 and 2026, global hyperscaler capex is now expected to grow +56% and +16% y/y respectively for AI training needs and stronger-than-expected customer demand, up from the pre-earnings expectation of +44% and +8%, according to Morgan Stanley. Hyperscaler 2025 capex expectations have increased by more than 50% from a year ago, according to Morgan Stanley. All hyperscalers are still supply constrained versus the AI demand they are seeing, so there is scope for further upgrades.
This constructive backdrop continues to feel broadly analogous with the mid-1990s, the last time the infrastructure for general purpose technology was built out in the form of the internet. While that earlier period proved positive for returns, it was also punctuated by bouts of volatility (seven 15%+ NASDAQ 100 drawdowns between 1995-98). With equity markets back at highs and valuations back at the high end of their post-global financial crisis trading range, there could be turbulence ahead. In our view, volatility is part and parcel of investing during the early stages of a new technology cycle. With a portfolio beta10 well above one and with active share versus the index at 50%, we continue to hold modest amounts of NASDAQ 100 put options11 to help ameliorate the impact of any potential market setbacks.
However, we continue to strongly believe that AI represents the next general purpose technology and that the investment opportunity has not been fully reflected by markets. Recent stock price travails at highly respected companies such as Gartner and the London Stock Exchange Group reflect a rapidly evolving investment narrative with investors increasingly unwilling to give companies the benefit of the doubt in the event of potential AI competition. We expect this trend to continue, if not intensify, reflecting our long-held view of an accelerating timeline to AI disruption. We are hopeful this should support a rich environment for active managers (single stock dispersion is elevated) and a meaningful opportunity for our 11-strong experienced team of technology investment specialists to continue to deliver differentiated returns to the benchmark and the peer group.
1. A trade-weighted currency measures a country's currency's value against a basket of currencies, adjusted for how much the country trades with each country; it shows that currency's overall strength globally
2. A type of cryptocurrency whose value is pegged to another asset (fiat currency; gold; precious metal etc)
3. A basis point is a common unit of measure for interest rates and other percentages in finance; one basis point equals 0.01%
4. The total value of recurring revenue expected from active customer contracts over a year
6. The largest cloud service providers (AWS; Microsoft Azure; Google Cloud; Meta Platforms; Apple; TikTok)
7. Earnings before interest, taxes, depreciation, and amortization; used to assess a company's profitability and financial performance
8. Stands for Gross Domestic Product and is the standard measure of the value created through the production of goods and services in a country
9. Apple, Microsoft, Alphabet, Amazon, NVIDIA, Meta Platforms and Tesla
10. A measure of a stock's volatility compared to the market/ an index; the market/index has a beta of 1 with each stock rated at +/-1 in comparison
11. A put option grants the right to the owner to sell some/all of an underlying security at a specified price, on or before the option's expiration date
