AI demand is no longer confined to the companies selling the chips. This quarter, it showed up in filings from 44 distinct companies, spanning semiconductors, communications, security and software. The common phrase was data center and AI demand, with the business consequences varying sharply by company.
The share of filers flagging the topic rose to 29.5%, from 16.4% in the same-length window last quarter. That is about a 1.80x quarter-over-quarter increase, a noticeable step up but not a corporate-wide conversion. Share matters here: 44 companies mentioned the topic this quarter versus 63 last quarter, but the denominator of all companies filing was different.
Marvell described the cleanest version of the theme, with demand reaching across several product lines.
"Strong revenue growth from our data center market was driven by AI-related demand for a broad range of our products, including electro-optics, custom, storage, and switching."
Marvell Technology / 10-Q / 2026-08-28
For Marvell, AI is not a single-product explanation. It is a reason given for growth across electro-optics, custom products, storage and switching.
Coherent put a number on the same appetite. Its Datacenter & Communications segment produced a large increase, with AI demand doing much of the explaining.
"Revenues increased $1,124 million, or 43%, in the Datacenter & Communications segment, with increases in datacom driven primarily by ongoing strong AI datacenter demand and growth in our telecom revenue due to higher demand in the data center interconnect and the telecom transport business."
Coherent / 10-K / 2026-08-14
That filing ties the theme to datacom, interconnect and telecom transport, not just to a server rack. The language is upbeat, and the reported change is concrete.
Fabrinet made the demand chain one step more visible by pointing to its customers rather than its own end market.
"This increase was primarily due to an increase in our key customers’ demand for both data center products and communications infrastructure products."
Fabrinet / 10-K / 2026-08-18
The point is less that every company is selling directly to data centers than that the spending is traveling through suppliers and infrastructure providers.
There is a less comfortable version of the theme in Napco’s filing. More demand can also mean more competition for the components needed to serve it.
"Increased demand for semiconductors and electronic components driven by artificial intelligence (\"AI\") infrastructure and data center expansion could adversely affect our supply chain and operating results."
Napco Security Technologies / 10-K / 2026-08-24
Napco’s wording is conditional, so it describes a sensitivity rather than a reported problem. Still, it shows how the same investment cycle can appear in a security company’s risk language.
OSI was more direct about what it had already experienced: tighter availability and higher costs for memory and semiconductor components.
"We have experienced tighter supply conditions and increased costs for certain memory associated and semiconductor components, reflecting a broader global imbalance between supply and demand for memory used in data center and AI related infrastructure."
OSI Systems / 10-K / 2026-08-21
That is the other side of the filing pattern. Most tagged mentions read as positive, with an approximate tally of 17 positive and two negative. But the negative language is specific: AI infrastructure can lift demand, while also putting pressure on the inputs required to build around it.
The earnings-season pattern is therefore not that every company suddenly sounds like a chipmaker. It is that data center and AI demand is becoming a shared explanation for growth, customer activity, infrastructure spending and, in a few filings, supply-chain exposure. A larger share of companies reached for the phrase this quarter. Their own words show why the phrase can mean both more orders and a tighter parts market.
This is a descriptive reading of SEC filings via jodie’s analytics, not investment advice.