Briefing After Hours | Weekly Market Recap (July 27–31, 2026): The Fed, Big Tech Earnings and the Semiconductor Reset
By Ieva Bružaitė | Founder, Orion Capital | August 2026
Executive Summary
Some weeks test portfolios. This one tested convictions. In five trading sessions, markets absorbed a Federal Reserve decision with the most divided committee in years, four Magnificent Seven earnings reports, record results from the world’s leading memory maker that were nonetheless sold, and a bond market pushing long-term yields to levels last seen in 2007. By Friday’s close, the verdict was clearer than the noise suggested: the S&P 500 and Dow each gained roughly 1% for the week, and the Nasdaq advanced about 1.6% — even as July as a whole ended nearly flat for the S&P 500 and down 3.2% for the Nasdaq.
The deeper story of the week was not any single print. It was the market learning to distinguish, in real time, between companies that spend on AI and companies whose spending already earns. Microsoft rose 15.5% in a single session — the largest one-day gain in its history. Meta, growing faster in percentage terms, fell roughly 10%. Same week, similar spending, opposite verdicts. That divergence, not the Fed, was the week’s true lesson.
| Index | Friday Close | Week | July |
| S&P 500 | 7,489.72 | +~1.0% | −0.1% |
| Nasdaq Composite | 25,373.85 | +~1.6% | −3.2% |
| Dow Jones | 52,485.03 | +~1.0% | +0.3% (4th monthly gain) |
| US 10-Year Yield | ~4.73% | Highest since Jan 2025 | — |
| US 30-Year Yield | ~5.25% | Highest since 2007 | — |
| WTI Crude | ~$85 / bbl | Higher on Hormuz risk | — |
The Federal Reserve: A Hold That Did Not Feel Like One
On Wednesday, the Federal Open Market Committee left its benchmark rate unchanged at 3.50–3.75%. On paper, a non-event. In substance, anything but. Three voting members — Beth Hammack, Neel Kashkari and Lorie Logan — dissented in favour of a quarter-point hike, a level of open disagreement the committee has rarely displayed in recent years. The statement acknowledged an economy “expanding at a solid pace” while conceding that inflation remains above the 2% target, pressured in part by energy prices as hostilities between the United States and Iran continue to threaten traffic through the Strait of Hormuz.
Chair Kevin Warsh’s press conference offered resolve without a map: the committee “will deliver price stability,” he said, but declined to specify the path. Markets read the ambiguity as hawkish. The Dow fell as much as 840 points on Wednesday afternoon, and the bond market delivered its own commentary through the week — the 10-year Treasury yield climbed above 4.7%, its highest level since January 2025, while the 30-year pushed toward 5.25%, territory untouched since 2007.
Our interpretation: with oil near $85 and Brent at $90, the Fed is caught between supply-driven inflation it cannot control and an equity market it does not wish to validate. Three dissents toward a hike mean the next move is genuinely uncertain — and for the first time in this cycle, uncertainty points in both directions. For long-term investors, the practical consequence sits in the long end of the curve: a 5.25% thirty-year yield quietly raises the bar every risk asset must clear.
Big Tech Earnings: The Market Grades the Bill
Four of the Magnificent Seven reported within thirty hours, and the results read like a controlled experiment on a single question: when does AI spending become AI earnings?
| Company | Headline Result | 2026 AI Capex Signal | Market Verdict |
| Microsoft | ~$90B revenue, +18% YoY; Azure +43%; Azure AI run rate $37B, +123% | ~$175B for calendar 2026; FY27 to grow further | +15.5% Thursday — largest one-day gain in company history |
| Meta | $60.8B revenue, +28% YoY; net income −14% to $15.85B | FY26 range raised to $130–145B | Roughly −10% — fastest growth, harshest verdict |
| Amazon | Profit beat; AWS +37% YoY; chip business expanding | Guidance still forming; spend rising | +15% Friday |
| Apple | Services and China revenue short; chip shortages raised costs | Modest by hyperscaler standards | −7% Friday despite growth |
Microsoft answered the question most cleanly. Roughly $90 billion in quarterly revenue, Azure growing 43%, and — the number that moved the stock — an AI business inside Azure running at a $37 billion annualised rate, up 123% year over year. Investors can now draw a straight line from capital expenditure to metered, external, growing revenue. Thursday’s 15.5% surge, the largest single-day gain in the company’s history, was the market paying for that legibility.
Amazon told the same story a day later: a profit beat, AWS accelerating 37%, and an expanding custom-chip business that shortens the distance between spending and monetisation. The stock jumped 15% on Friday.
Meta grew revenue 28% — faster than Microsoft — and was punished with a roughly 10% decline, because profits fell 14% as the company lifted its 2026 capital expenditure range to $130–145 billion. The market’s message was not that Meta is spending too much; Microsoft spends more. It was that Meta’s AI spending still lacks an external revenue line investors can measure. Apple, meanwhile, fell 7% after Services and China revenue disappointed and chip shortages raised costs — a reminder that in a supply-constrained cycle, even the world’s most valuable company queues like everyone else.
Add the four hyperscalers together and 2026 capital spending guidance now totals roughly $720–745 billion. A year ago, the debate was whether such numbers were possible. This week’s debate was narrower and healthier: which of these dollars already earn, and which are still promises.
Semiconductors: Record Profits, Sold First, Understood Later
No episode captured the week’s psychology better than SK Hynix. On Wednesday, the memory leader reported the strongest quarter in its history: revenue of ₩79.3 trillion, up 257% year over year; operating profit of ₩60.5 trillion, up 557%, at a 76% operating margin; and first-half revenue crossing ₩100 trillion for the first time. The stock fell as much as 15% before closing 9.6% lower in Seoul.
The reason was not the results but the ruler. Consensus had expected ₩84 trillion in revenue and ₩64 trillion in operating profit; part of the shortfall traced to HBM4 revenue recognition sliding into later quarters as mass shipments, which began in Q2, ramp through the second half. When a company misses expectations while growing profit sixfold, the problem is not the business — it is the altitude of the expectations. Notably, by Friday the shares had recovered sharply, and the company’s guidance quietly strengthened the structural case: DRAM demand seen up mid-twenties percent this year, multi-year contracts signed with roughly ten customers, and capacity expansion explicitly tied to confirmed demand rather than hope.
The rest of the complex followed the same arc: heavy selling into Wednesday’s Fed decision and the SK Hynix print, then a sharp reversal after Microsoft’s results reframed the demand question. Nvidia, AMD, Micron and their peers rallied Thursday and Friday, helped by Samsung’s Thursday report hinting at prolonged chip shortages and by Korean July semiconductor exports surging 179%. The week ended where the month began: with supply tight, demand rising, and the argument confined to price.
Sentiment Versus Fundamentals
It is worth stating plainly what this week demonstrated. Sentiment moved violently: a six-day Nasdaq losing streak, a 17% collapse in the VIX on Thursday, single stocks swinging 15% in a session. Fundamentals barely moved at all: cloud growth accelerated, memory capacity remained sold out, hyperscaler capex guidance rose, and long-term contracts lengthened. When prices travel that much further than facts, the movement is telling you about positioning and patience — not about the underlying businesses. The investor’s task in such weeks is unglamorous: verify the facts, ignore the tone of voice in which they are reported.
Orion Capital Perspective
A week like this rewards a simple discipline: separate the three questions markets kept blending together. Is the AI buildout real? The $720–745 billion answer arrived in writing. Is it profitable? Increasingly, and unevenly — the market now pays a visible premium for spending it can audit, which is why Microsoft and Amazon were rewarded while Meta and Apple were not. Is it fairly priced? That is the only question still open, and rising long-term yields have made it harder, not easier, to answer generously.
Our view remains what it was a week ago, only with more evidence: the correction of late July was a repricing of expectations, not a repudiation of demand. The bond market has raised the toll on the bridge to the future; it has not closed the bridge. In that environment, the businesses that matter are those whose customers must buy — the memory nobody can substitute, the cloud capacity already contracted — rather than those whose customers merely might.
What We Watch Next Week
- US payrolls and inflation expectations — the first major data the divided Fed will see after its three-dissent meeting, and the clearest input into whether the hawks gain a fourth vote.
- Oil and the Strait of Hormuz — with WTI near $85 and Brent at $90, energy is now the main channel through which geopolitics reaches the inflation data.
- The long end of the curve — a 30-year yield above 5.25% is the highest hurdle rate equities have faced in nearly two decades; whether it stabilises matters more than any single earnings report.
- Nvidia’s supply chain reads — after Samsung’s shortage commentary and SK Hynix’s HBM4 ramp, monthly Taiwan and Korea export data become a real-time proxy for AI demand.
- Remaining earnings — semiconductor equipment makers and enterprise software will show whether AI spending is broadening beyond the hyperscalers.
Conclusion
July closed with the S&P 500 essentially flat, the Nasdaq down 3.2%, and the Dow quietly recording a fourth consecutive monthly gain — three indices telling three different stories about the same economy. The week’s real dividing line ran elsewhere: between spending and earning, between record results and recorded expectations, between the reactive clock and the anticipatory one. The companies that crossed that line were rewarded within hours. For everyone else, the market’s message was patient but firm: the AI revolution no longer pays for promises. It pays for receipts.
Disclaimer: This publication is independent research and reflects the views of the author. It is provided for informational purposes only and does not constitute investment advice or a recommendation to buy or sell any security. Investors should conduct their own analysis before making decisions.
References
- Federal Reserve → https://www.federalreserve.gov
- Microsoft Investor Relations → https://www.microsoft.com/investor
- SK hynix investor Relations → https://www.skhynix.com
- Nasdaq Market Data → https://nasdaq.com
About the Author
Ieva Bružaitė is the Founder of Orion Capital and an independent market researcher focused on artificial intelligence, semiconductor infrastructure and long-term investing. Her research focuses on AI infrastructure, semiconductors, data centres and the companies shaping the next generation of technology. Her work combines independent market research with long-term investment analysis for technology-focused investors.
Read more about Ieva Bružaitė and Orion Capital→