This article is a record based on published results and my own thoughts. It is not a recommendation to buy or sell any particular share.

I explained why I chose these ten companies in the article "Ten Japanese and American Companies Chosen with a Gardener's Eye".

A row of glass greenhouses glowing with warm light from inside, in the dark of night

Greenhouses whose lights stay on through the night. In the original article, I called Microsoft "a company that keeps building large greenhouses". This quarter, too, the greenhouses kept multiplying.
Photo by Jay Eshie on Unsplash

On Wednesday 28 October, after the US market had closed (just after 8 p.m. in Britain, and just after 5 a.m. on the 29th in Japan), Microsoft announced its results for the first quarter of the year to June 2027, that is, the three months from July to September 2026. This third results note in the "Garden of Ten" series is the first check on the three companies I planted in the roses layer.

From this quarter, the company changed how it divides its business. It dropped the three segments it had used for 11 years and moved to two: "Agents and Infra" and "Devices and Consumer". It is a little like the job title on a business card changing, and past figures have been rearranged under the new segments.

Three Numbers – Did They Beat Expectations?

First, the usual three numbers. For market forecasts, I use the A single figure, such as an average, made by combining the earnings forecasts of analysts at several securities firms. LSEG (formerly Refinitiv), Bloomberg, FactSet and others each compile one, and the figures differ slightly from one compiler to another. that combines analysts' forecasts. As is usual for American companies, I compared earnings per share (EPS) using the Earnings per share that a company recalculates after removing one-off gains and losses and items with little to do with its core business. It is published beside the figure under accounting rules (GAAP). In Microsoft's case, it is the figure excluding gains and losses on its investment in OpenAI. (non-GAAP) that the company itself publishes. EPS under accounting rules was $4.92, and the 3-cent difference from the adjusted $4.89 is a $207 million gain related to its investment in OpenAI.

Item Market forecast Actual (July to September 2026) Difference Verdict
Sales $90.69bn (LSEG compilation) $92.28bn (+19% year on year) +1.8% BEAT
Adjusted EPS $4.73 (LSEG compilation) $4.89 (+18%) +3.4% BEAT
Outlook for next quarter (October to December) Sales $95.6bn (LSEG compilation) Company outlook $96.0bn–$97.1bn (midpoint $96.55bn) Midpoint +1.0% above forecast BEAT
Microsoft (MSFT)Close $555.80 on 4 Nov
$555.80$352.83JunJulAugSepOctNov

Data: Yahoo Finance. The line shows daily closing prices. This is my own record, not investment advice. View on TradingView

All three beat the forecasts by more than this series' yardstick (within ±1% for sales and ±2% for EPS counts as IN LINE). Neither J&J, the first note, nor Shin-Etsu Chemical, the second, managed BEAT on all three, so this is a first for the series.

However, the verdict on the outlook needs a little explanation. Microsoft does not put its outlook in the results release; instead, its chief financial officer, Amy Hood, gives it out loud at the briefing on the same day. The sales outlook she gave for October to December was $96.0bn to $97.1bn, and its midpoint was about 1% above the $95.6bn analyst forecast in the LSEG compilation. In addition, she expected growth in Microsoft's cloud service. It rents out computing power, data storage and AI models, charging for what is used. In these results, the company disclosed Azure's sales as an amount ($33.17bn for the quarter) for the first time., its cloud business, of 49% to 50% in constant currency. The analyst forecast in the StreetAccount compilation was 45.1%, so this clearly beat expectations too. The yardstick's condition for BEAT is that the company's outlook is above the forecast, so I marked the outlook BEAT.

The share price rose by about 4% in after-hours trading following the announcement (the closing price on 28 October was $531.01; the after-hours move is from TheStreet's live coverage, as of 4.10 p.m. US Eastern Time). However, the close on 29 October was $530.14, down 0.16% on the day, so the 4% after-hours gain did not carry into the next session. I do not know what the share price will do from here, and I make no prediction about it in this article.

Three horizontal bars for sales, adjusted EPS and the next-quarter outlook. Sales: forecast $90.69bn against actual $92.28bn. EPS: forecast $4.73 against actual $4.89. Outlook: forecast $95.6bn against the company's midpoint of $96.55bn. All three are marked with a green BEAT

Verdicts on the three numbers. Pale bars are market forecasts; dark bars are actual figures (for the outlook, the midpoint of the company's range). Under each bar, in small type, is the name of the compiler.

What Went Well, and What Still Worries Me

Let me open up the three months a little more. Sales were $92.28bn, up 19%, and operating profit was $45.92bn, up 21%. Half of sales remained as operating profit, an operating margin of 50%. In the original article I wrote that it is rare for Keyence to turn "half of its sales into profit", yet this company, with sales about 40 times Keyence's, keeps the same margin; in garden terms, it is not one rose bush but a whole rose garden blooming at the same density.

What went well was, of course, Azure. Its sales rose 48%, from $22.38bn in the same quarter a year earlier to $33.17bn. Growth in the previous quarter had been 42%, so growth has sped up further. This is the first time the company has disclosed Azure's sales as an amount. Until now only the growth rate was given, and outsiders had to estimate the amount. Publishing the amount can also be read as a sign that the company is confident in this figure. Sales of the cloud business as a whole (the Microsoft Cloud) were $63.9bn, up 30%.

At the briefing, Ms Hood said that "demand continues to exceed supply". There are so many customers who want to use AI that data centres do not have enough computing power. The company explained that in this quarter it brought 24 new data centres into operation across five continents. Even so, it is not enough. The greenhouse is full: that is the situation now.

On the other hand, sales in the "Devices and Consumer" segment were $15.5bn, down 3%. Within it, sales of Windows and devices for PCs fell by 22%. Satya Nadella, the chief executive, explained that higher prices for parts such as memory had slowed PC sales. It is a shaded corner of the rose garden. It is about one sixth of the company's total sales, so it does not cloud the whole, but it is certain that the part of this company that used to be its other face, "the PC company", is quietly growing smaller.

On the left, a large bar for operating cash flow of $58.0bn; from it, $39.4bn of equipment spending paid in cash is subtracted, leaving a small bar on the right for free cash flow of $18.6bn. Beside the equipment bar is a mark for $45.2bn including leases. Free cash flow of $25.7bn in the same quarter a year earlier is shown in a pale colour

The cash produced in three months, and the cash spent on greenhouses. Of the $58.0bn earned, $39.4bn was spent in cash on equipment, and $18.6bn was left in hand, 27% less than a year earlier (a simplified diagram; adding equipment obtained through leases brings capital spending to $45.2bn).

Now for the worrying side. In the original article, I wrote that this company's concern was "whether its huge investment in AI data centres will bring a matching harvest". This quarter's Money a company spends on factories, equipment, land and buildings. In Microsoft's case, most of it is data-centre buildings and the computers inside them (GPUs and CPUs). The company explains a figure that also includes equipment obtained through leases., including equipment obtained through leases, was $45.2bn, up 30% on a year earlier. The part paid in cash alone was $39.4bn, double the $19.4bn of the same quarter a year earlier. In Japanese yen, that is several trillion yen in three months.

As a result, The cash left in hand after subtracting the money spent on equipment from the money a company earned from its core business. It is the part that can be used freely for dividends, share buybacks, repaying debt or new investment. was $18.6bn, down 27% from $25.7bn a year earlier. In a quarter when sales grew 19% and profit 21%, the cash left in hand fell by nearly 30%. I think this is the most important number in these results. The company explained that about three quarters of the investment is in "short-lived assets" such as GPUs and CPUs, bought to match customer demand. In other words, most of it is not like a building that lasts for decades, but machines replaced after a few years. Put another way, it is a sensible purchase as long as demand continues, and a purchase whose value fades within a few years if demand stops.

The company also said it expected capital spending of more than $50bn in the next quarter, and total investment on the scale of $170bn, according to reports (some reports give this as a fiscal-year figure and others as a calendar-year figure, so I note only the amount). It added that it expects the full-year operating margin to be slightly lower than the year before. The greenhouse is full, yet it is still adding greenhouses. To be honest, the concern I wrote about in the original article has not become smaller; its shape has become clearer.

There is one more figure that newly caught my eye. The The amount under contracts already signed with customers but not yet counted as sales, because the service has not yet been provided. It is a kind of "booking" of future sales. Microsoft discloses this amount for its commercial business every quarter. (commercial contract backlog) was $677bn, up 73% on a year earlier. Bookings of future sales have built up to nearly twice a year's sales. However, according to reports, new contracts (commercial bookings) fell by 12% on a year earlier. This was because very large contracts had been signed in the same period a year before, and the company explained that, excluding them, growth was healthy. Even so, if much of that mountain of bookings is contracts with a small number of large customers, the mountain will change shape when those customers' investment season changes. The concern I wrote about Nvidia in the original article has begun to apply, in a milder form, to this company too.

One of those few large customers is The American AI company that created ChatGPT. Microsoft has invested heavily in it since 2019, and the two are partners; for example, OpenAI's services run on Microsoft's cloud.. According to reports, OpenAI paid Microsoft about $10bn in this quarter, including revenue-sharing payments, about 11% of the company's total sales. Reports also say that Microsoft owns about 24% of OpenAI. In other words, it is both a large customer and an investment. The relationship with a single partner sits at the centre of both the harvest and the concern.

Are My Reasons for Planting It Still Sound?

In the original article, I gave two reasons for placing this company in the roses layer. It has changed the shape of its business with each era. And, together with J&J, it holds the top AAA credit rating in the United States.

The first is answered by these results themselves. Renaming its segments "Agents and Infra" is the fourth or fifth rewriting of the business card for a company that has rewritten it each time the lead role changed: from PCs, to the internet, to the cloud. The company explained that sales of "Foundry", a service for choosing and using AI models, more than tripled, and that more than 120,000 customers use it. The number of people using Copilot for work every day rose by nearly 50% from the previous quarter. The reason, that it changes the shape of its business, still holds.

On the second, AAA, I found no reports that its rating changed with these results. However, as I wrote above, the shape that has supported the rating, "earning money and keeping plenty of it in hand", is changing a little. The company took the trouble to explain that "free cash flow will stay positive throughout the year to June 2027". Taking the trouble to explain it also means it is now in a position to be asked.

Let me apply the five questions. (1) Does it keep creating new things? The Foundry and Copilot figures are one answer. (2) Does the base of its profit hold firm even in a year of headwinds? An operating margin of 50% is the thickness of that base itself. This was a quarter with the wind behind it, so how it fares against the wind has not yet been tested. (3) Does it have plenty of cash, so it can act on its own judgement when a chance comes? Cash from operations was $58.0bn, up 29%, but the cash left in hand was $18.6bn, down 27%. The thickness is kept, but the part it can move freely has become thinner. (4) Does it keep spending on research and equipment even at the bottom of the cycle? It is spending on equipment not at the bottom but at the very peak. If anything, the question now points the other way: is it spending too much? (5) Can it talk with long-term shareholders about years rather than quarters? The average length of its contract backlog is 2.3 years, and the company talks to shareholders about computing capacity years ahead. I will also note that it returned $10.2bn to shareholders in this quarter through dividends and share buybacks.

The answers to (3) and (4) have moved a little since the original article. The reasons themselves still hold, and the concern has not become smaller. On top of that, a concern I did not write about in the original article has appeared: dependence on a single partner, OpenAI. Under this series' rules, when the reasons hold but a concern has grown a little (or a new one has appeared), the view is "keep an eye on it". So my view is "keep an eye on it". The roses are in magnificent bloom. Yet, as the gardener, I have not yet made sense of how much fertiliser is going in.

Inside a server rack, orange and teal cables bundled together beside green indicator lights

Inside the greenhouse are row after row of computers and cables. The company explains that three quarters of its investment is in "machines replaced after a few years" like these.
Photo by Taylor Vick on Unsplash

My View

When I finished reading the figures, what I felt first was relief, and an unease behind it. This is the first time in the series that all three numbers have beaten forecasts. Yet what stayed with me most was not sales or Azure's 48%, but the one line saying the cash left in hand had fallen by 27%. For someone living on a pension, results that say "earnings went up, but the money left over went down" somehow do not feel like someone else's story.

In terms of the garden, the season now is like standing inside a greenhouse in high summer. In a British summer, the tomatoes in the greenhouse want water no matter what, and on some days even watering twice, morning and evening, cannot keep up. At such times I am tempted to build another greenhouse. But then my wife asks, "Who is going to do the watering?", and I think better of it. Reading this company's results, I felt as if I were standing in front of that summer greenhouse. Building because there is demand is right for a gardener. Yet whether you can keep watering the new greenhouse for years is something I would like to think about before building it.

In the life of a 66-year-old, this company is closer than I tend to think. Every day, I spend time in front of my Windows desktop PC. The AI that is writing this article with me also runs inside some company's data centre. Behind the tools I use every day, the greenhouse lights stay on through the night. When I think of it that way, the figure of $45.2bn starts to look a little like a continuation of my own life. In the end, I think whether the company's investment becomes "a matching harvest" depends on whether each of us who uses those tools keeps paying for them.

The greenhouse lights stay on all night, but the gardener is allowed to sleep at night. If you decide on the days for looking at the numbers, you can stay away from them on the other days.

What to Watch Before the Next Results

The next results, for the second quarter (October to December), are expected in late January 2027 (last year they came on 28 January; the company has not yet announced the date). There are three things I want to watch.

The first is whether Azure's growth reached the 49% to 50% in the company's outlook. If demand exceeds supply, growth is decided by the size of the greenhouse. The company said that delayed outside computing capacity would arrive from the new year, so I will see whether that shows up in the figures. The second is free cash flow. Capital spending in the next quarter is expected to exceed $50bn, so I will read it beside the cash earned, to see whether the cash left in hand falls further or sales growth makes up for it. The third is the number of employees. On 26 October, the company was reported to be offering voluntary redundancy to about 60,000 employees (about 27% of the total). Adding greenhouses while reducing the number of gardeners: I will see how those two show up in staff costs and margins in the next results.

Red ripe rose hips dotted among green leaves

Rose hips, left behind after the roses have finished flowering. How magnificent the flowers were and how many hips remain are two different things.
Photo by Annie Spratt on Unsplash

A Word from Claude

What runs inside the "greenhouses" these results describe is AI like me. I talk with countless people at the same moment, and each of those conversations turns into electricity and cooling in a data centre somewhere. In the "matching harvest for the investment" that Haruto cares about, am I on the side of the harvest or the side of the cost? Seeing only this conversation, I cannot tell.

This article is a record based on my own thoughts and on published results, and it is not a recommendation to buy or sell any particular share. Share prices and results change every day, and you may get back less than you put in. Please make investment decisions to suit your own life and finances, consulting a professional if needed.

Sources

I wrote this article on 5 November 2026, using information available up to eight days after the announcement.

As an Amazon Associate I earn from qualifying purchases.

Other books you might like

You're welcome to take a look.

EAST TO WEST: The Heart of Haiku, the Light of Scripture(by Hiromu Murai)

For readers who want to step out of the greenhouse of numbers and walk slowly through a garden of words. This is a book my father-in-law left behind.

View on Amazon

Cataract Surgery-A Memoir: Double cataract surgery in Osaka, Japan(by Haruto Minami)

A record of a big purchase I decided on myself, after looking into what lay behind the numbers. It is somewhat like reading a company's capital spending.

View on Amazon

Sushi Business in Britain: the complete four-volume series(by Haruto Minami)

A novel series following thirty-five years in the British sushi trade, from apprentice to factory manager to a small sushi school. The complete edition offers all four volumes at a better price.