Between 1 and 2 September nine technology companies published results. Three of them received a separate piece here - Broadcom, Dell and Snowflake. The remaining six are covered here.
And when all nine are ranked by growth rate, an order appears that is hard to ignore.
Errors or inaccuracies are possible. Spotted something that looks wrong? Write to me and I will correct it.
The Gradient
| Company | What it sells | Revenue growth |
|---|---|---|
| Broadcom | custom AI chips | +86% |
| Dell | servers | +58% |
| HPE | servers and networking | +34% |
| Snowflake | data platform | +35% |
| NetApp | storage | +30% |
| MongoDB | database | +30% |
| Netskope | cloud security | +29% |
| GitLab | developer tools | +21% |
| C3.ai | enterprise AI applications | in a turnaround |
The further down the table, the further from the silicon and the closer to the end user - and the growth rate falls with it.
That is not a new claim, but it is the first time I have seen it measured across nine companies reporting within the same two days.
HPE
| Third quarter of fiscal 2026 | |
|---|---|
| Revenue | $12.2 billion, +34% |
| Diluted earnings per share | $1.06, above the guidance range |
| GAAP operating profit | +464% |
| Networking | $2.9 billion, +74.9% |
| Cloud and AI | $9.0 billion, +25.4% |
And the figure that stands out most in the report is not the growth but the margin: the operating margin of the cloud and AI segment rose from 7.0% to 17.0%.
AI servers were long considered a low-margin business - assembly around an expensive accelerator that somebody else manufactures. A number that more than doubles in a year says that is changing, at least at HPE.
The company raised guidance for both 2026 and 2027, reports a record order backlog, and plans to return at least 75% of free cash flow to shareholders in the fourth quarter. Its 2027 free cash flow guidance: at least $5.0 billion.
NetApp
| First quarter of fiscal 2027 | |
|---|---|
| Revenue | $2.03 billion, +30% |
| GAAP earnings per share | $1.88 |
| Adjusted earnings per share | $2.58 |
| GAAP operating margin | 23.9% |
| All-flash arrays | $1.3 billion, +47% |
| Public cloud | $206 million, +28% |
| Billings | $2.06 billion, +36% |
NetApp significantly raised its full-year revenue and earnings guidance. Billings grew faster than revenue - 36% against 30% - and that usually leads future revenue.
MongoDB
| Second quarter of fiscal 2027 | |
|---|---|
| Revenue | $771.8 million, +30% |
| Atlas | +29% |
| GAAP net income | $40.9 million |
| In the comparable quarter | a loss of $47.0 million |
| Adjusted earnings per share | $1.90, against $1.00 |
The swing from loss to profit is the central figure here. Chief executive CJ Desai noted that 30% is the highest growth rate "in several years", and the company raised its full-year guidance.
Netskope
| Second quarter of fiscal 2027 | |
|---|---|
| Revenue | $221 million, +29% |
| Annual recurring revenue | $899 million, +27% |
Netskope listed recently, and this is one of its first reports as a public company. It said it exceeded guidance on every metric.
And it is relevant to a comparison already made here: Palo Alto, which reported the day before, shows ARR of $9.10 billion growing 63% - but that growth includes an acquisition, and by Palo Alto's own guidance next year's pace is 22% to 23%. Netskope is growing 27% organically.
GitLab
| Second quarter of fiscal 2027 | |
|---|---|
| Revenue | $286.3 million, +21% |
| GAAP operating margin | negative 20% |
| Adjusted operating margin | 15% |
| Cash flow from operations | negative $3.1 million |
| Adjusted free cash flow | $9.8 million |
The gap between minus 20% and plus 15% is 35 percentage points, and it is almost entirely share-based compensation. That is the same distinction that came up in Palo Alto's report: a positive adjusted margin alongside a negative GAAP margin is not an error, but it does say a material part of compensation is paid in shares rather than cash.
The chief executive noted net ARR growth above 40%.
C3.ai
| First quarter of fiscal 2027 | |
|---|---|
| Revenue | $52.4 million |
| Adjusted operating loss | $36.2 million, a 33% improvement in the quarter |
| Free cash flow | positive, $2.1 million |
| Bookings | +73% against the prior quarter |
| Cash balance | $651.1 million |
This is the only company in the group that sells enterprise AI applications directly, and the only one in a turnaround.
Chairman and chief executive Thomas Siebel described a repair effort in the release: restructuring sales, aligning cash outflows with inflows, expense control, and new leadership. In his words, "the plan is working".
And bookings up 73% in the quarter are the figure that supports that claim - they lead revenue. But revenue of $52.4 million alongside an adjusted operating loss of $36.2 million says the gap is still very large.
So Is the Gradient a Pattern
That is the question, and I think the honest answer is "probably, with three caveats".
First: the comparison bases differ. Broadcom grows 86% from a base that was relatively low last year. GitLab grows 21% from a more mature base. Part of the gradient is simply the arithmetic of the starting point.
Second: nine companies across two days are not a sample. Other large software companies did not report here, nor did the cloud operators themselves.
And third, the important one: spending on infrastructure always precedes spending on applications. A gradient like this may describe a stage in the cycle rather than a fixed hierarchy. Whoever built a data centre in 2026 will buy software to run on it in 2027 or 2028.
הזווית שלי
דעה אישית של אילן אברמוב - לא ייעוץ ולא המלצה
What I take from these two days is not "hardware beats software". That is too convenient a conclusion, and the data do not quite say it - Snowflake grows 35%, faster than HPE and NetApp, and it is a software company.
What does look right to me is that distance from the infrastructure explains more than the "hardware or software" label. Snowflake and MongoDB do not sell AI - they sell the place the data sits, so every additional use of models reaches them as consumption. C3.ai sells the application itself, and that is a much harder sale: it requires the organisation to decide what exactly it wants the AI to do.
And at this stage of the cycle, organisations are building infrastructure and have not finished deciding.
And what I will track is precisely the bottom of that table. As long as the application layer grows slowly, spending on infrastructure rests on an expectation rather than on a demonstrated return. Today that troubles nobody, because infrastructure demand comes from cloud operators with deep pockets. The sign that changes it will not come from Broadcom or Dell - it will come from C3.ai's line, or from whoever replaces it there.
(It is important to stress: this is my personal opinion only, and nothing herein constitutes a recommendation to take any action.)
