Below you will find my full portfolio as of today 27 August 2026:

Read more

S&P 500 −1.43%
Nasdaq 100 −2.45%
STOXX Europe 600 −0.56%
EURO STOXX 50 −1.18%
My portfolio –1.51%

Market pulse

Global equities declined during the week ending 21 August as rising government-bond yields and higher oil prices outweighed resilient earnings and improving German manufacturing data. The bond sell-off was the central driver: expanding fiscal deficits, heavy sovereign issuance and persistent inflation pushed long-term borrowing costs to multi-year highs, producing particularly heavy selling in semiconductors and other valuation-sensitive technology shares. Iran-related uncertainty compounded the pressure by lifting energy prices and strengthening expectations that both the Federal Reserve and ECB would maintain restrictive policy. In Europe, the DAX fell 1.15%, from 26,440.31 to 26,136.56, with German technology and industrial companies affected by higher yields, producer-price inflation and low Rhine water levels. Strong manufacturing data produced a partial rebound on Friday, but not enough to offset the deterioration in financial conditions earlier in the week.

Crowd vs. price

The dominant longer-term market narrative remained constructive: corporate earnings were strong, AI investment continued to expand and several major research houses still expected further equity gains. Actual price action, however, showed that investors were unwilling to ignore the cost of financing that growth. All four tracked indices declined, the Nasdaq 100 underperformed markedly, and defensive healthcare and energy shares held up better than high-duration technology companies.

The week therefore looked less like the beginning of an earnings recession and more like a valuation and positioning correction. Investors did not abandon the AI or economic-growth narratives, but they demanded lower prices as bond yields rose. Friday’s rebound, together with continued strength in selected healthcare, energy and consumer shares, showed that the selling remained selective rather than indiscriminate. The market’s message was that earnings remained credible, but the price investors were prepared to pay for distant growth had fallen.

Holdings & Watchlist Notes

Marvell Technology — MRVL: +6.77%

Marvell Technology gained 6.77% during the week, driven overwhelmingly by its expanded custom-silicon agreement with Google. The arrangement creates a potentially exceptional long-term revenue opportunity and validates Marvell as a major supplier of hyperscaler-designed AI infrastructure. It also suggests that Google is diversifying beyond its traditional custom-chip partners, improving Marvell’s strategic position in one of the semiconductor industry’s fastest-growing markets. The announcement landed immediately after a broad chip-sector sell-off, magnifying the rebound. Potential dilution from the accompanying Google warrant and short-term profit-taking limited the final weekly gain, but the commercial and strategic implications remained decisively positive.

Read more

S&P 500: +0.36%
NASDAQ 100: +1.09%
STOXX Europe 600: −0.36%
EURO STOXX 50: +0.24%
My portfolio: – 0.18%

Market pulse

The dominant equity-market driver was a conflict between two opposing forces: softer US inflation and consumption data reduced the expected probability of further Federal Reserve tightening, while renewed tension around the Strait of Hormuz restored the geopolitical premium in oil prices. Brent crude gained 6.0% over the week and WTI 5.4% as US–Iran negotiations stalled, tanker attacks resumed and the United States threatened to maintain its naval blockade indefinitely. At the same time, July consumer and producer inflation came in sufficiently mild for markets to reduce the implied probability of a September Federal Reserve increase to approximately 30%, from around 50% one week earlier. The result was not a broad risk-on rally, but a narrow US advance led by memory, AI infrastructure and rate-sensitive technology shares.

Monday and Tuesday were both negative sessions. On Monday, fading expectations of an agreement over the Strait of Hormuz pushed US crude approximately 5% higher to $82.13 a barrel. The S&P 500 declined 0.06%, while the NASDAQ 100 fell 0.34%. Intel also lost 4.1% after announcing a $15 billion share sale, and NVIDIA declined 2.9%. On Tuesday, the S&P 500 fell another 0.32% and the NASDAQ 100 0.33% as Iran said the strait would remain closed unless the United States accepted its conditions. Amazon declined 2.1% and Alphabet 3.8%, adding company-specific pressure to the technology indices. By Tuesday’s close, the S&P 500 was approximately 0.38% below the previous Friday’s level and the NASDAQ 100 approximately 0.66% lower.

Wednesday and Thursday produced the entire net advance for the week. US consumer prices increased only 0.1% in July, while annual inflation eased from 3.5% to 3.4%. Core inflation was 0.2% month-on-month and 2.5% year-on-year, down from 2.6%. The following day, headline producer prices were unchanged, compared with expectations for an increase. Markets interpreted the releases as evidence that the Federal Reserve could remain on hold in September. The S&P 500 gained 0.26% on Wednesday and 0.65% on Thursday, reaching a record close, while the NASDAQ 100 rose 0.74% and 1.15% respectively.

The inflation reports were market-friendly, but not uniformly benign beneath the headlines. Producer prices excluding food, energy and trade services increased 0.4% in July and 4.7% year-on-year, while consumer energy prices remained 14.7% above their level one year earlier despite falling during July itself. The market nevertheless concentrated on the lower headline figures and declining probability of an immediate rate increase. This was not, however, a conventional lower-yield rally across the full maturity curve: the two-year Treasury yield declined by approximately three basis points during the week, but the ten-year yield rose approximately four basis points. Investors therefore priced out some near-term Federal Reserve risk without concluding that longer-term inflation and fiscal risks had disappeared.

Corporate results supplied the fundamental support behind the technology rebound. CoreWeave rose 19% after beating quarterly expectations and increasing its annual capital-expenditure forecast. Nebius gained 34%, while other data-centre operators also advanced. On Thursday, Sandisk rose 13.7% after presenting ambitious long-term growth and profitability targets, pulling Micron 4.2% higher and strengthening the argument that AI-related memory demand could remain structurally tighter than in previous semiconductor cycles. Approximately 85% of the S&P 500 companies that had reported results had exceeded earnings expectations, while earnings were estimated to have risen 32.7% excluding mark-to-market gains at Alphabet and Amazon.

The rally nevertheless remained highly selective. Cisco declined sharply even after forecasting annual revenue above consensus because its gross-margin guidance disappointed and its AI growth was concentrated in relatively hardware-intensive sales. Applied Materials fell on Friday despite results that exceeded expectations, demonstrating that objectively good numbers were no longer enough when valuations already assumed strong AI demand. The market rewarded companies that could demonstrate scarcity, pricing power or an identifiable physical bottleneck, while punishing companies whose growth came with weaker margins or insufficient upside relative to expectations.

Friday’s US consumption data introduced a different risk. Retail sales unexpectedly declined 0.6% in July, their first fall in nine months and largest decline in 14 months. The economically important control group fell 0.4%, compared with expectations for a 0.3% increase, while preliminary University of Michigan consumer sentiment dropped from 55.2 to 51.0. Some of the decline reflected temporary factors—including Amazon moving Prime Day into June, falling gasoline receipts and the fading effect of unusually large tax refunds—but economists nevertheless reduced third-quarter growth estimates. The S&P 500 fell 0.17% and the NASDAQ 100 0.13% on Friday. Weak consumption lowered the probability of a rate increase, but it also began to move the market from “bad news is good news” toward the less favourable question of whether household demand was genuinely losing momentum.

European equities were caught between strong earnings and renewed energy risk. The STOXX Europe 600 declined 0.36%, ending a four-week winning streak, while the more concentrated EURO STOXX 50 gained 0.24%. Aggregate second-quarter earnings for STOXX 600 companies were expected to rise 23.4%, their strongest growth in almost four years, and 58.6% of reporting companies had exceeded expectations. Excluding energy, earnings growth was still estimated at 12.3%. Europe’s dependence on imported energy nevertheless made the rise in oil prices more damaging than it was for the United States, while the region had less exposure to the memory and AI-infrastructure companies responsible for the US technology advance.

The clean reading of the week is that it was a lower-Fed-risk but higher-oil-risk market. Softer inflation and consumer data supported the valuation of long-duration growth companies, but the return of the Hormuz premium prevented a broader cyclical rally and weighed particularly on Europe, Sweden and transport companies. The NASDAQ 100’s outperformance did not represent indiscriminate enthusiasm for technology. It was driven primarily by a small group of companies positioned at bottlenecks in memory, computing capacity and AI infrastructure.

Crowd vs. price

Investor positioning shifted forcefully back toward memory and selected AI-infrastructure companies. The immediate catalyst for the Korean rally was a report that Singapore’s Temasek was considering additional investment in Samsung and SK Hynix. Temasek subsequently clarified that it had not consulted the Korean government about the timing of any investment and had already owned both companies for more than two years, meaning that a new transaction was not confirmed. The size of the reaction therefore indicates that short-covering and position rebuilding amplified the headline. Sandisk’s long-term targets then provided a more fundamental justification for the wider memory-sector rerating.

The week further clarified the hierarchy within the AI trade. Memory producers, cloud-capacity providers and companies exposed to identifiable infrastructure constraints were rewarded. Cisco, by contrast, was punished because high AI order growth came with a hardware-heavy mix and weaker gross-margin guidance. Applied Materials was sold after its report because results that would previously have been considered strong failed to exceed already elevated expectations by a sufficient margin. Datadog’s rebound demonstrated that investors were willing to reverse an excessive selloff, but not that its customer-concentration problem had disappeared. The dividing line remained visible present demand and strategic scarcity versus valuations that required nearly flawless future execution.

Holdings & Watchlist Notes

Samsung Electronics

+18.83%

Samsung Electronics gained 18.83%, making it the week’s strongest stock in the measured portfolio list. The most immediate catalyst was a Korean media report that Temasek was considering an investment in Samsung and SK Hynix. Both shares rose more than 7% on Wednesday and helped lift the KOSPI by more than 4%. The report should not be treated as a confirmed new transaction: Temasek said it had not sought advice from the Korean government about investment timing and had already invested in both companies more than two years earlier. The rally nevertheless continued as investors returned to Korean memory shares and reassessed the durability of AI-related demand. Expectations of stronger shareholder distributions supplied an additional catalyst after Samsung said that it was examining sustainable methods of enhancing returns and expected to provide details soon. The clean reading is that a speculative headline triggered a large positioning reversal, which was then reinforced by memory-sector fundamentals and the prospect of increased capital distributions. The scale of the move was substantially greater than the amount of confirmed new company information.

Read more

Below you will find my full portfolio as of today 14 August 2026:

Read more

S&P 500: +3.58%

NASDAQ 100: +5.12%

STOXX Europe 600: +1.70%

EURO STOXX 50: +2.61%

My portfolio: + 3.27%

Market pulse

The dominant equity-market driver was the simultaneous removal of two risks that had constrained valuations during July: the geopolitical premium in oil prices and the expectation of further Federal Reserve tightening. Hopes that negotiations could lead to a reopening of the Strait of Hormuz pushed crude oil and long-term Treasury yields sharply lower, while strong corporate earnings—particularly among semiconductor, AI-infrastructure and technology companies—provided a fundamental reason for investors to re-enter risk assets. The result was a broad rally, but one led decisively by the NASDAQ 100, which gained 5.12% and substantially outperformed all five other benchmarks.

Monday and Tuesday produced most of the week’s advance. On Monday, crude oil fell approximately 5% as the market anticipated negotiations over the Strait of Hormuz. Lower oil reduced near-term inflation expectations and pulled Treasury yields lower, helping the S&P 500 rise 1.48% and the NASDAQ Composite 2.13%. On Tuesday, another decline of roughly 5% in crude was accompanied by further optimism regarding a possible agreement, sending the S&P 500 up 1.79% and the NASDAQ Composite up 2.59%. Together, those two sessions accounted for approximately 92% of the S&P 500’s full-week gain. The underlying mechanism was straightforward: cheaper energy reduced pressure on consumers and corporate margins, while simultaneously lowering the probability that the Federal Reserve would need to raise interest rates again.

The rally was reinforced by unusually strong corporate results and renewed confidence that AI capital expenditure was creating revenue beyond the largest hyperscalers. Palantir rose 29.5% after raising its annual revenue forecast, while Caterpillar gained 5.6% as AI data-centre construction increased demand for its power-generation and construction equipment. Semiconductor shares rose particularly sharply, with the Philadelphia Semiconductor Index gaining 6.6% on Tuesday. By Friday, 85.1% of the 436 S&P 500 companies that had reported results had exceeded earnings expectations. The market therefore moved from questioning the entire AI investment cycle to rewarding companies able to demonstrate that the build-out was producing identifiable demand for chips, memory, networking, power and physical infrastructure.

Friday’s US employment report produced the final leg of the rally. Nonfarm payrolls unexpectedly fell by 23,000 in July, compared with expectations for an increase of 80,000, while the previous two months were revised sharply lower. The unemployment rate nevertheless declined to 4.1%, largely because workers left the labour force. Markets interpreted the report primarily through the interest-rate channel: the implied probability of a September Federal Reserve increase fell to approximately 44%, from 55% the previous day and 67% one week earlier. The S&P 500 consequently gained another 0.62% and the NASDAQ Composite 1.30%. This was a conventional “bad news is good news” reaction, although the deterioration in employment also created a risk that future weak data could eventually be interpreted as evidence of economic contraction rather than monetary relief.

European equities participated in the rally but were less exposed to the high-beta technology rebound. The STOXX Europe 600 recorded four consecutive record closes and finished the week at 660.25, supported by lower oil prices, improving US rate expectations and a broadly positive European reporting season. Technology was the index’s best-performing sector, while estimated second-quarter European profit growth rose above 22%, its strongest pace since 2022. The EURO STOXX 50 outperformed the broader STOXX Europe 600, while both Swedish indices gained approximately 2%. Europe benefited particularly from the decline in crude because its economies are generally more dependent on imported energy than the United States.

The advance was not indiscriminate. Several technology companies fell despite publishing objectively strong results because their guidance failed to meet expectations already embedded in their valuations. Datadog was the clearest example: the company reported 36% revenue growth, but investors focused on reduced usage by a major AI customer and a full-year forecast below the prevailing analyst consensus. AMD and several software companies also came under pressure during the middle of the week. The market was therefore not abandoning valuation discipline altogether; it was rewarding companies whose exposure to AI demand appeared immediate and independently verifiable, while punishing those dependent on flawless future execution or a small number of unusually large customers.

The clean reading of the week is therefore that it was a lower-oil, lower-yield rally, validated by earnings and extended by weak employment data. The market’s strongest preference was for semiconductor, memory, optical-connectivity and data-centre infrastructure companies. The principal question was no longer whether AI expenditure was excessive in the aggregate, but which companies controlled the physical bottlenecks through which that expenditure had to pass.

Crowd vs. price

Investor positioning moved rapidly back toward high-beta AI and semiconductor shares. Credo, Marvell, Intel, NVIDIA, ASML, Micron and Applied Materials all substantially outperformed the broader market. After the previous month’s semiconductor correction, stronger earnings and falling yields gave underexposed investors a reason to rebuild positions. The speed and concentration of the rebound suggest that short-covering and position rebuilding amplified the fundamental catalysts, although the precise contribution of those flows cannot be measured from public price data alone.

The week nevertheless exposed a more demanding hierarchy within the AI trade. Credo and Marvell were rewarded for exposure to memory, optical and high-speed connectivity bottlenecks. Intel benefited from improving operational results and renewed interest in higher-beta semiconductor turnaround stories. Datadog, by contrast, was punished even though current revenue growth remained strong because its forward outlook exposed dependence on the usage decisions of one major AI customer. The dividing line was not AI versus non-AI. It was visible demand and strategic scarcity versus valuation dependent on future expectations remaining almost perfect.

The two utility holdings demonstrated that company-specific events could still override the positive macroeconomic environment. NextEra Energy and Dominion Energy both declined after Virginia’s governor intervened in the regulatory review of their proposed merger. Their weakness was therefore not evidence of a general collapse in defensive equities; it principally reflected a reassessment of the transaction’s timing, conditions and implied value.

Holdings & Watchlist Notes

Credo Technology (CRDO)

+20.73%

Credo Technology gained 20.73%, although the increase cannot be attributed to a single earnings report or financial disclosure. The most defensible explanation is a combination of product news, favourable policy read-through and the broad semiconductor rally. During FMS 2026, Credo presented new connectivity products designed to move data more efficiently between processors, memory and storage, including its OmniConnect Weaver memory interconnect and a PCIe 6.0 retimer for next-generation AI infrastructure. The company therefore benefited from growing recognition that data movement and memory access—not merely processor availability—are becoming major constraints on AI-system performance. Sentiment was reinforced by reports that the US government was preparing restrictions on imports of new Chinese optical transceivers used in data centres. Credo supplies high-speed electrical and optical-connectivity technology and could benefit if US customers are pushed toward non-Chinese suppliers, although that policy connection remains an investor inference rather than company guidance. The clean reading is that Credo was revalued as a potential beneficiary of both AI-connectivity demand and the possible exclusion of Chinese competitors, with the sector-wide semiconductor rebound magnifying the movement.

Read more

Below you will find my full portfolio as of today 04 August 2026:

Read more

S&P 500: +1.05%

NASDAQ 100: +0.52%

STOXX Europe 600: +0.73%

EURO STOXX 50: +1.23%

My portfolio: -1.37%

Market pulse

The dominant equity-market driver was another change in how investors interpreted the AI capital-expenditure cycle. The previous reporting period had ended with investors punishing hyperscalers for negative free cash flow and rapidly rising infrastructure expenditure. This week, Microsoft and Amazon demonstrated that the market will tolerate enormous spending when it is accompanied by visible cloud growth, margins and contracted demand. Microsoft rose more than 15% on Thursday after strong Azure growth and a forecast implying continued cash generation, while Amazon jumped 15.3% on Friday after AWS revenue increased 37%—its fastest growth in more than four years. The two reports transformed the late-week narrative from indiscriminate fear of AI expenditure into a more selective test of whether each company can prove near-term monetisation.

That distinction produced extreme divergence inside Big Tech. Amazon and Microsoft were rewarded; Meta remained under pressure after reporting a 91% decline in free cash flow, and Apple fell 7.4% on Friday after issuing a weak outlook linked to supply constraints and concern about demand at higher prices. The market was not embracing every company spending heavily on AI. It was rewarding the companies that could connect that spending to accelerating revenue and operating performance.

The rally also had to overcome a significant monetary-policy shock. The Federal Reserve kept the federal-funds target at 3.50%–3.75% on Wednesday, but the decision passed by a divided 9–3 vote, with three policymakers preferring an immediate increase. The lack of clear forward guidance from Chair Kevin Warsh increased uncertainty over September and helped push long-dated Treasury yields to multi-year highs. The S&P 500 fell 1.52% on Wednesday and the NASDAQ 100 2.06%. Strong Microsoft and Amazon results subsequently outweighed that pressure at index level, but higher discount rates remained an important restraint on expensive growth shares.

Semiconductors experienced the most violent version of this conflict. Before the Microsoft report, investors sold the sector on concerns about stretched valuations, the financing of AI infrastructure and stronger Chinese competition. Samsung Electronics fell 13.4% and SK Hynix 14.7% on Tuesday, while China’s CXMT surged 466% on its Shanghai debut and reports of domestic Chinese deep-ultraviolet lithography equipment raised fears of faster capacity expansion. Micron, AMD, KLA, ASML and other chip-related holdings were pulled into the same de-risking wave. Microsoft’s results then drove the Philadelphia Semiconductor Index up 8.2% on Thursday, but the rebound was insufficient to erase the earlier damage for many individual holdings.

European equities were more broadly supported by corporate earnings. Estimated second-quarter STOXX 600 profit growth rose to 20.8% year-on-year; even excluding the energy sector, the estimate was 10.3%. Strong results across consumer, financial, industrial and energy companies helped the STOXX Europe 600 reach an intraday record on Friday, although it closed slightly lower that day. The EURO STOXX 50 outperformed both the broader European index and the NASDAQ 100 over the full period. The two Swedish indices performed best of the six benchmarks, with both gaining approximately 1.5%.

Oil remained an important but unstable macroeconomic variable. Brent crude fell sharply early in the week as hopes of U.S.–Iran de-escalation reduced the immediate supply premium, but renewed tension later pushed it briefly above $90 a barrel. Lower oil supported travel, consumer and rate-sensitive shares; higher oil benefited energy producers but revived inflation and interest-rate fears. The week’s swings in crude and Treasury yields repeatedly changed the relative performance of technology, energy, airlines and consumer stocks.

The market’s message therefore evolved beyond the simple division between companies receiving AI capital expenditure and companies funding it. Amazon and Microsoft showed that the funders can outperform when monetisation is immediate and measurable. KLA, Micron and AMD showed that suppliers can fall even while current demand remains strong when valuations imply perfect execution and investors become more concerned about competition, financing and future oversupply. The new dividing line is evidence versus expectation.

Crowd vs. price

Investor attention remained concentrated in AI, but the week exposed a more demanding hierarchy inside the trade. Amazon and Alphabet gained because investors became more confident that cloud and AI infrastructure could produce rapid revenue growth. Micron and AMD fell because the hardware side of the trade was forced to absorb concerns about Chinese competition, customer financing and valuations that already assumed years of exceptional growth.

The same AI investment cycle therefore generated both the portfolio’s three largest gains and its three largest losses. The crowd did not abandon AI. It moved toward companies offering visible near-term monetisation and away from companies whose prices required future demand, margins and competitive positions to remain almost flawless.

Holdings & Watchlist Notes

Amazon (AMZN)

+17.00%

Almost the entire weekly gain occurred on Friday after Amazon reported its strongest AWS growth in more than four years. AWS revenue increased 37% to $42.2 billion, while the contract backlog reached $496 billion. Amazon simultaneously increased its expected 2026 capital expenditure by 10% to $220 billion. Under ordinary circumstances, that spending increase could have damaged the shares; instead, management said that most available AWS capacity for 2027 and some capacity for 2028 had already been reserved by customers. Investors therefore treated the expenditure as construction against identifiable demand rather than a speculative build-out.

The principal caveat is that trailing twelve-month free cash flow deteriorated from positive $18.2 billion to negative $7.6 billion. The rally was not evidence that the cost problem had disappeared. It reflected a judgment that the AWS acceleration and reservation data were strong enough to justify the cash burn. At least 15 brokerages raised their price targets after the report. The clean read is that Amazon gained because it supplied the clearest evidence of the week that enormous AI expenditure can already generate measurable revenue, margins and contracted demand.

Read more

S&P 500: −0.61%
NASDAQ 100: −1.62%
My portfolio: + 0.1%

Market pulse

U.S. equities finished the period lower, with technology again underperforming the broader market. The S&P 500 declined from 7,457.69 to 7,411.98, a loss of 0.61%, while the NASDAQ 100 fell from 28,592.66 to 28,128.34, a decline of 1.62%. The technology-heavy index therefore underperformed the S&P 500 by 1.01 percentage points. The stock portfolio declined by approximately 0.5% in SEK, roughly matching the S&P 500 and outperforming the NASDAQ 100 by about 1.1 percentage points. The portfolio received some currency protection from the dollar, which appreciated approximately 0.64% against the Swedish krona during the period.

The dominant market driver was a change in how investors interpreted the AI capital-expenditure cycle. Alphabet reported its first negative quarterly free cash flow, burning $5.9 billion despite Google Cloud revenue increasing 82%. The company simultaneously increased its expected 2026 capital expenditure by $15 billion. Tesla also reported negative free cash flow and sharply higher investment requirements. Alphabet subsequently fell approximately 7% on Thursday and Tesla 14.5%, while Meta and Amazon declined in sympathy. The market was no longer asking only whether AI demand was strong. It began asking whether revenue could grow quickly enough to exceed capital expenditure, depreciation, financing and operating costs.

The technology shock coincided with another increase in geopolitical and inflation risk. Brent crude settled above $100 a barrel on Thursday and U.S. crude above $92 as fighting involving the United States, Iran and the Houthis threatened shipping through both the Strait of Hormuz and the Red Sea. The ten-year Treasury yield approached 4.7%, its highest level since January 2025. Higher oil prices threatened household spending and corporate margins while also reducing the Federal Reserve’s ability to lower interest rates. The combination of weaker technology sentiment, higher energy costs and rising yields produced Thursday’s broad selloff.

The final weekly losses nevertheless conceal a powerful early-week semiconductor rebound. On Tuesday, the Philadelphia Semiconductor Index rose 5.2% as investors bought back shares following the previous week’s sharp correction. Micron gained 12.2%, and other memory, storage and AI-infrastructure companies rallied strongly. The buying reflected both oversold conditions and fear that investors had reduced their semiconductor exposure immediately before potentially strong technology earnings.

That rebound partially reversed on Friday. Intel fell 7.9%, the semiconductor index declined 4.5% and the NASDAQ Composite lost another 0.64%, even though the S&P 500 edged 0.05% higher. Oil fell nearly 4%, providing some support to consumer, real-estate and other rate-sensitive shares, but investors continued selling semiconductor and AI-related positions. The same hyperscaler spending that had supported chip demand earlier in the week was now being interpreted as possible overinvestment by the companies financing it.

The broader market was considerably more resilient than the headline technology selloff suggested. Energy shares gained approximately 3.8% over the week, while industrials, materials and real estate also advanced. This sector rotation explains much of the S&P 500’s outperformance of the NASDAQ 100. New U.S. tariffs of 10% or 12.5% on goods from 60 trading partners added a secondary inflation and margin risk, but tariffs were not the principal cause of the week’s decline. Earnings, AI spending, oil and Treasury yields were more immediate drivers.

The market’s message was therefore not that AI demand had disappeared. Demand remained exceptionally strong. The market instead divided the AI value chain into companies that receive capital expenditure and companies that must fund it. Memory, accelerators and networking suppliers initially benefited from evidence that infrastructure construction remained intense. Alphabet, Meta, Amazon and other platform companies were punished because that same construction reduced free cash flow and made the eventual return on investment less certain.

Crowd vs. price

Investor attention remained concentrated in AI, but the week exposed a significant split inside the trade. Micron, AMD rose because greater hyperscaler expenditure translates directly into demand for memory, accelerators, servers and high-speed connectivity. Alphabet, Meta and Amazon fell because greater expenditure translates into lower free cash flow, higher depreciation and greater execution risk. The same data-centre project can therefore be positive for the supplier and negative for the customer financing it.

Holdings & Watchlist Notes

Micron Technology

+8.48%

The largest part of the gain occurred on Tuesday, when the shares rose 12.2% as the semiconductor index rebounded 5.2%. Micron had been hit particularly hard during the previous correction, making it one of the clearest candidates for a mean-reversion rally once investors began restoring semiconductor exposure. Continued hyperscaler construction also reinforced expectations of strong demand for high-bandwidth memory and conventional DRAM.

There was no new Micron earnings report during the period. The movement was primarily a repricing of existing memory-demand expectations rather than a response to newly disclosed company results. The shares surrendered a substantial part of their gain on Friday when investors again questioned the sustainability of AI infrastructure spending. The clean read is that Micron rose because the preceding selloff had become excessive relative to the continuing strength of AI-related memory demand, but the stock remained highly sensitive to changes in the market’s interpretation of hyperscaler capital expenditure.

Read more

Below you will find my full portfolio as of today 23 July 2026:

Read more

Weekly financial report

Reporting period: 10 July close to 17 July close

S&P 500: −1.55%
NASDAQ 100: −4.13%
My portfolio: approximately −4.35%

Market pulse

U.S. equities finished the period lower, with technology decisively losing market leadership. The S&P 500 declined 1.55%, while the technology-heavy NASDAQ 100 fell 4.13%, underperforming the broader index by 2.58 percentage points. The portfolio declined by an estimated 3.25%, placing it between the two indexes: its substantial exposure to semiconductors and AI infrastructure hurt performance, while gains in CrowdStrike, Cloudflare and Visa provided a partial offset.

The dominant market driver was a violent de-rating of semiconductor and AI-related shares. The Philadelphia Semiconductor Index fell approximately 10%, its largest weekly decline in more than a year, and finished more than 20% below its late-June record. The correction reflected profit-taking after an extraordinary previous-quarter rally, growing scrutiny of the returns generated by AI capital expenditure and the unwinding of leveraged momentum positions. U.S. growth funds experienced $7.18 billion of net withdrawals, while value funds received $3 billion, confirming that the move extended beyond isolated company news into a broader portfolio rotation.

The market was not rejecting the existence of strong AI demand. TSMC reported a record 77% increase in quarterly profit and indicated that capital expenditure would remain exceptionally high, yet its shares still declined. This demonstrated that reported fundamentals were no longer sufficient when valuations already assumed near-perfect demand. Sentiment was further weakened by Moonshot AI’s introduction of Kimi K3, which raised questions about the competitive threat from lower-cost Chinese models, and reports that Google was behind schedule with Gemini 3.5 Pro. The central concern shifted from whether AI usage would grow to whether the enormous amount of infrastructure spending would ultimately produce adequate financial returns.

Geopolitical risk compounded the technology correction. Renewed fighting between the United States and Iran and continuing disruption around the Strait of Hormuz pushed oil prices higher, reviving fears that energy costs would feed back into inflation, Treasury yields and Federal Reserve policy. Energy was the only S&P 500 sector to rise on Friday, helping the broader index outperform the NASDAQ 100, while airlines, travel companies and other fuel-sensitive businesses came under pressure.

There were meaningful positive counterweights. June CPI declined 0.4% month over month, while core CPI was unchanged, and producer prices fell 0.3%. These reports reduced the immediate case for another interest-rate increase and supported temporary equity rebounds on Tuesday and Wednesday. Strong early bank earnings also indicated that corporate activity, trading and dealmaking remained resilient. However, the inflation data described conditions before the latest rise in oil prices, limiting the durability of the relief.

The market’s internal message was therefore negative for crowded trades but not uniformly negative for equities. Of the first 49 S&P 500 companies to report, 90% exceeded earnings expectations, and aggregate second-quarter profit growth was projected at approximately 26%. Nevertheless, the indexes fell because positioning, valuation and index concentration outweighed otherwise strong aggregate earnings. Investors remained willing to pay for visible demand and identifiable monetization, but became much less tolerant of companies whose valuations required uninterrupted momentum and flawless execution.

Crowd vs. price

Investor attention remained concentrated in AI, but crowding itself became the principal risk. Semiconductor shares had risen almost 88% during the preceding quarter, leaving many companies highly vulnerable to profit-taking. Strong operational results from TSMC and other industry leaders failed to clear the market’s elevated expectations, while leveraged products and momentum strategies amplified the decline. The companies that had benefited most from the AI rally became the easiest positions for investors to sell when risk appetite weakened.

At the same time, the market rewarded companies with immediate evidence of demand or concrete new revenue opportunities. CrowdStrike and Cloudflare benefited from greater urgency around enterprise cybersecurity, while Visa advanced after introducing products designed to turn stablecoins and generative AI into additional services for financial institutions. The clean message was not that investors had abandoned AI. They were rotating away from speculative, capital-intensive AI exposure and toward companies where demand, pricing power or monetization could already be identified.

Holdings & Watchlist Notes

CrowdStrike (CRWD)

+8.49%

CrowdStrike was the strongest-performing holding over the period. The largest identifiable catalyst was IBM’s preliminary second-quarter warning on 14 July. IBM said customers had been distracted by rapidly evolving cybersecurity concerns even as they reprioritized technology budgets. What was negative for IBM was interpreted as positive evidence for specialist cybersecurity vendors: security remained urgent and comparatively difficult for enterprises to postpone. CrowdStrike and other cybersecurity shares rallied sharply following the disclosure.

CrowdStrike also expanded its strategic partnership with Schwarz Digits on 16 July. The agreement will bring the Falcon platform to European customers through the STACKIT sovereign cloud, while CrowdStrike agreed to acquire XM Cyber’s intellectual property, including attack-path visualization and exposure-management technology. The clean read is that CrowdStrike rose because IBM’s warning strengthened the case for durable cybersecurity spending, while the Schwarz Digits agreement added a concrete European distribution and product-expansion catalyst.

Read more

Below you will find my full portfolio as of today 14 July 2026:

Read more

S&P 500: +1.23%
NASDAQ 100: +1.69%
My portfolio: +0.53%

Market pulse

U.S. equities finished the period higher, with technology regaining market leadership despite considerable volatility beneath the index surface. Unlike the previous period, the technology-heavy NASDAQ 100 outperformed the broader market.

The most important market driver was the continuing debate over AI investment and semiconductor valuations. The period began with a strong rebound in AI-related shares, helped by Broadcom’s extended supply relationship with Apple. That recovery was interrupted on 7 July, when Samsung’s exceptionally strong results failed to meet the market’s elevated expectations. This triggered another sharp semiconductor selloff as investors questioned whether the enormous amount of capital being invested in AI infrastructure would generate sufficient returns.

The AI trade subsequently recovered. Nvidia benefited from reports that China could permit selected domestic AI companies to purchase limited quantities of its H200 processors. Broadcom continued to receive support from its Apple relationship, while Micron rebounded after presenting plans for more than $250 billion of U.S. investment through 2035. SK Hynix’s successful U.S. listing and Meta’s strong rally further reinforced demand for selected AI-related companies toward the end of the period.

Geopolitical and monetary-policy concerns nevertheless constrained the rally. Renewed tension between the United States and Iran pushed oil prices and Treasury yields higher, increasing the risk that energy costs could keep inflation elevated. The FOMC minutes also showed that Federal Reserve officials remained concerned about inflation, preventing investors from fully dismissing the possibility of another rate increase.

The market’s internal message was therefore mixed but constructive. Investors were still prepared to buy AI exposure, but only where there was visible evidence of demand, monetization or strategic value. Companies perceived as highly valued, excessively crowded or dependent on future expectations were vulnerable to severe corrections.

Crowd vs. price

Investor attention remained concentrated in AI, but became increasingly selective. The market moved sharply between valuation anxiety and renewed confidence in structural AI demand.

The market rewarded companies with concrete evidence of AI demand, differentiated technology, recurring revenue or identifiable monetization opportunities. It punished companies whose valuations appeared to require near-perfect execution or whose recent rallies had already incorporated much of the expected future growth. This explains how several prominent AI-related companies could move in opposite directions during the same period.

Holdings & Watchlist Notes

Meta Platforms (META)

+14.81%

Meta was the strongest-performing holding over the period. The rally was supported by renewed confidence in the company’s ability to convert its enormous AI investment into both product improvements and additional revenue opportunities. Investors responded positively to reports that Meta could rent out surplus computing capacity, reducing concerns that its infrastructure spending would remain entirely dependent on advertising returns.

The company also benefited from enthusiasm around its expanding AI capacity, custom-chip development and generative-AI products. These developments encouraged investors to view Meta’s capital expenditure as a potentially monetizable technological asset rather than simply a cost burden. The clean read is that Meta rose because the market became more confident that its AI investment could strengthen its core platforms while also creating new revenue streams.

Read more

Below you will find my full portfolio as of today 6 July 2026:

Read more

Performance

S&P 500: +1.76%
NASDAQ 100: +0.72%
My portfolio: +0.98%

Market pulse

U.S. equities finished higher in a holiday-shortened trading window, but the strength was uneven. The S&P 500 rose 1.76%, materially outperforming the NASDAQ 100’s 0.72% gain. The market opened the period with a relief rally after the previous week’s AI and technology selloff, helped by quarter-end positioning and a rebound in mega-cap technology. But the rally lost momentum on 1–2 July as the semiconductor and AI-hardware trade sold off again. The result was a positive week for the broader market, but a much more mixed one underneath the index surface. Index closes moved from 7,354.02 to 7,483.24 for the S&P 500 and from 29,118.24 to 29,329.21 for the NASDAQ 100.

The most important macro catalyst was the softer June employment report. Nonfarm payrolls rose by only 57,000, unemployment held at 4.2%, and prior months were revised lower. That reduced immediate Fed-hike fears and helped broader risk appetite, especially in blue-chip, defensive, and rate-sensitive parts of the market. But the benefit did not flow evenly into the most crowded AI hardware names. On 2 July, Reuters reported that the semiconductor index fell 5.4% for a second straight day, while the Nasdaq declined even as the Dow reached a record.

The market’s internal message was clear: investors were still willing to buy AI exposure, but they became more selective about where. Software, cybersecurity, enterprise AI, and selected blue-chip names performed well. Crowded semiconductor, memory, server, and AI-infrastructure names were punished. The S&P 500 held up better because breadth improved outside chips, while the NASDAQ 100 lagged because of its heavier exposure to semiconductors and mega-cap technology.

Crowd vs. price

Investor attention remained concentrated in AI, but price action became more discriminating. The market rewarded companies with software-like margins, visible recurring revenue, and immediate AI-adjacent demand.

Holdings & Watchlist Notes

CrowdStrike (CRWD)

+10.67%

CrowdStrike was one of the strongest U.S.-listed holdings, rising more than 10% over the period. The move was driven by a combination of bullish analyst commentary, strong platform-security demand, and positioning around its 4-for-1 stock split, which began trading on a split-adjusted basis on 2 July. Wells Fargo’s checks pointed to continued enterprise spending on platform-based cybersecurity, while CrowdStrike’s recent results showed strong ARR growth, record net new ARR, and continued momentum in AI-security products. The clean read is that CrowdStrike worked because investors treated it as a high-quality AI-security infrastructure company rather than a generic software name.

Read more

Performance

S&P 500: −1.95%
NASDAQ 100: −4.24%
My portfolio: −4.26%

Market pulse

U.S. equities finished lower in a holiday-shortened trading window, with the NASDAQ-100 falling 4.24% and materially underperforming the S&P 500’s 1.95% decline. The main driver was a sharp unwind in the crowded AI and semiconductor trade. Investors moved from rewarding AI exposure to questioning the cost, financing, and timing of AI infrastructure returns. The pressure was most visible on 23 June, when the semiconductor index fell sharply as investors scrutinized debt-funded AI spending and priced in a more hawkish Federal Reserve path.

The selloff was not evenly spread across the market. High-valuation AI infrastructure names, semiconductors, and OpenAI-adjacent stocks were hit hardest, while healthcare, software, and some defensive or enterprise-technology names held up better. On 26 June, the S&P 500 ended only slightly lower, but AI-related chip stocks sold off again, and Reuters reported that the chip index fell 5.3% that day and 7.9% for the week. Healthcare was one of the few clear havens, helped by strong moves in names such as Eli Lilly and Moderna.

The result was a sharp reversal from the prior week’s risk-on tape. Last week, investors rewarded AI scarcity and memory pricing power; this week, they punished capital intensity, debt financing, and long-duration AI payoffs. The NASDAQ-100 underperformed because it has heavier exposure to mega-cap technology, semiconductors, and AI infrastructure. The S&P 500 held up better because defensive sectors and healthcare cushioned the broader index.

Crowd vs. price

Investor attention remained concentrated in AI, but price action became far less forgiving. The market stopped treating all AI exposure as positive and started separating companies with immediate earnings power from companies whose AI upside depends on massive upfront spending, debt financing, and future monetization.

Holdings & Watchlist Notes

Eli Lilly (LLY)

+9.97%

Eli Lilly was one of the strongest holdings in the portfolio, rising almost 10% as investors rotated into healthcare and rewarded company-specific catalysts. The biggest driver was renewed confidence in Lilly’s obesity-drug franchise after the Medicare GLP-1 Bridge program improved the access story for drugs such as Zepbound and Foundayo. The rally was reinforced by positive European regulatory news: Lilly announced that the EMA’s CHMP recommended Jaypirca for adults with chronic lymphocytic leukemia across all lines of therapy. The broader market backdrop also helped, because healthcare became a haven as investors moved out of crowded technology and semiconductor names. The clean read is that Lilly went up because investors saw both defensive growth and renewed upside in its GLP-1 pipeline.

Read more

Performance

S&P 500: +0.93%
NASDAQ 100: +2.60%
My portfolio: +5.17

Market pulse

Between the June 12 and June 18 closes, U.S. equities finished higher in a holiday-shortened trading window, with the NASDAQ-100 gaining 2.60% and materially outperforming the S&P 500’s 0.93% gain. The dominant macro driver was relief around the U.S.–Iran interim peace agreement and the reopening of the Strait of Hormuz, which pushed oil lower, reduced inflation fears, and supported risk assets. Reuters described the June 15 move as a classic relief rally, with major indexes and bonds rising while oil fell to a three-month low.

The week was not purely risk-on, though. The market sold off sharply on June 17 after the Federal Reserve signaled that its next move could be a rate hike, putting pressure on long-duration growth and software names. The rebound on June 18 was concentrated in semiconductors, with the Philadelphia Semiconductor Index up 6.4%, helped by Intel’s Apple-related rally and renewed enthusiasm for AI infrastructure. The result was a selective tape: memory, AI compute, semiconductors, and scarce growth assets outperformed, while overextended software, IT-services, and legacy-networking names were punished.

Crowd vs. price

Investor attention remained concentrated in AI, but the market became more discriminating. The winners were companies where investors saw scarcity, pricing power, or forced demand.

Holdings & Watchlist Notes

Micron Technology — MU

+15.52%

Micron rose 15.52%, from $981.61 to $1,133.99, driven by a sharp re-rating of the AI memory cycle. The biggest immediate catalyst was a wave of analyst target increases ahead of earnings, with investors increasingly treating Micron as a scarce AI-infrastructure supplier rather than a normal cyclical memory company. The rally was reinforced by Apple’s warning that rising memory and storage costs were becoming hard to absorb, which validated the view that memory suppliers still had pricing power. Underneath the move, Micron’s recent numbers already supported the thesis: record revenue, very high gross margins, sharply higher DRAM and NAND pricing, and guidance for another record quarter. The clean read is that investors bought Micron because AI-driven demand had turned memory into a high-margin bottleneck, and the June headlines suggested the upcycle was still accelerating.

Read more

Below you will find my full portfolio as of today 16 June 2026:

Read more

Performance

S&P 500: +0.65%
NASDAQ 100: +2.34%
My portfolio: –0,74%

Market pulse

Between the June 5 and June 12 closes, U.S. equities finished higher despite a sharp midweek drawdown. The NASDAQ-100 gained 2.34%, materially outperforming the S&P 500’s 0.65% gain, because the strongest rebounds were concentrated in AI, semiconductors, memory, and server-CPU names. The dominant macro driver was the interaction between U.S.–Iran war risk, oil prices, inflation, and Fed expectations. Escalation headlines and hotter inflation data pressured stocks around June 10, while the market rebounded sharply after Trump called off planned strikes, oil prices fell, and investors began pricing lower geopolitical and inflation risk. Underneath the index move, the tape remained highly selective: Intel, Micron, and Arm were rewarded as AI-infrastructure beneficiaries, while Super Micro Computer, Adobe, and Oracle were punished where AI growth came with dilution, margin pressure, capex intensity, or weaker monetization. The move was not a rejection of AI infrastructure; it was a sharper sorting of winners and losers inside the AI trade.

Crowd vs. price

Investor attention remained concentrated in AI, but the market became more discriminating. The winners were companies where investors saw scarcity, validation, or improving strategic positioning.

Holdings & Watchlist Notes

Intel — INTC:

+25.61%

Intel was the strongest performer of the period, rising 25.61% from the June 5 close to the June 12 close. The main catalyst was a reported Google order for more than 3 million TPUs to be manufactured by Intel in 2028, which gave investors a concrete hyperscaler datapoint for Intel Foundry. That was followed by Bank of America’s rare double upgrade from Underperform to Buy and a price-target increase to $135, reinforcing the idea that Intel could benefit from both AI server CPU demand and the industry’s need for advanced manufacturing capacity outside TSMC. The rally was amplified by Intel’s depressed starting point after the previous semiconductor rout. The clean interpretation is that Intel was re-rated as a potential AI-infrastructure and foundry comeback story, though the rally still carries execution risk because much of the enthusiasm rests on future customer wins rather than near-term revenue.

Read more

Performance

S&P 500: −2.59%
NASDAQ 100: −4.53%

Bitcoin: −17.83%

My portfolio: –5,4%

Market pulse

Between the May 29 and June 5 closes, U.S. equities reversed sharply from record-high momentum into a concentrated risk-off selloff. The NASDAQ-100 fell 4.53%, materially underperforming the S&P 500’s 2.59% decline, because the damage was concentrated in AI, technology, and semiconductor stocks. The week began with continued enthusiasm around AI infrastructure, but that optimism cracked after Broadcom’s results failed to meet the market’s elevated AI-chip expectations. The decisive macro blow came on June 5, when the May jobs report showed 172,000 payrolls added and unemployment unchanged at 4.3%, pushing investors to price in a more hawkish Federal Reserve path and reducing hopes for rate cuts. Broadcom’s disappointing AI guidance, rising rate fears, Middle East/oil-related inflation concerns, and stretched positioning after a nine-week rally combined to trigger a violent valuation reset across crowded tech and chip names. The move was not a rejection of the AI infrastructure theme, but a sharp repricing of expectations: investors demanded more proof, better visibility, and lower tolerance for anything short of perfection.

Crowd vs. price

Investor attention remained concentrated in AI infrastructure, but the market became far more selective. Marvell was the clear outlier: it rose because investors saw it as a validated AI-infrastructure winner after Nvidia CEO Jensen Huang publicly described it as a potential “next trillion-dollar company,” reinforcing confidence in Marvell’s custom-chip, networking, and data-center role. By contrast, Qualcomm, Broadcom, and Intel were hit by different versions of the same problem: expectations had become too high, valuations were stretched, and investors started punishing companies where the AI story looked competitive, delayed, or not strong enough to justify the prior rally. Bitcoin was hit even harder because crypto faced both macro pressure and its own internal selling cycle: ETF outflows, forced liquidations, weaker speculative appetite, and capital rotating toward AI and large tech-related opportunities.

Holdings & Watchlist Notes

Marvell Technology (MRVL): +28.52%

Marvell was the standout performer of the period, rising 28.52% from the May 29 close to the June 5 close. The main catalyst was Nvidia CEO Jensen Huang’s Computex endorsement, where he described Marvell as a potential “next trillion-dollar company.” That comment mattered because it validated Marvell’s position inside the AI infrastructure ecosystem, especially in custom chips, networking, optics, and data-center interconnect. The rally was also supported by Marvell’s own growth outlook: the company has forecast that its custom-chip business could exceed $10 billion in revenue by fiscal 2029, while its latest quarterly results showed record revenue and strong AI-related data-center demand. The stock did fall sharply on June 5 during the broader semiconductor selloff, but the early-week AI rerating was large enough to leave Marvell strongly positive for the full period.

Read more

Below you will find my full portfolio as of today 2 June 2026:

Read more