Earnings reports are always important, both to assess a company’s position and to validate (or invalidate) the market assumptions they’re making in their product planning and positioning. This quarter, the third calendar quarter of 2025, will be particularly important because of the massive bets tech companies have placed on AI. Obviously, companies like Google and Microsoft will be scrutinized for their AI results, but so will most of the major network companies, like Cisco and HPE/Juniper, because they’ve been pushing an AI-centric story to investors. Any major signs of weakness will likely be punished, both by investors and by the financial and tech media.
In networking, this is also the quarter we can expect to see the early signals of the new shape of the Cisco/Juniper rivalry, created when HPE bought Juniper. The move created the first true unified server/network vendor since IBM left the business, and since HPE was second only to IBM in strategic account influence, it created the first major rival to Cisco as well. The Gemini Deep Research report I posted on the network market made this same remark; HPE/Juniper was its most significant network market event.
What can we expect from the reports? First, I don’t think the third quarter represents a point where AI shifts like the Gartner “trough of disillusionment” really hit. There’s still a lot of data-center AI buildout happening among the giants, and the OpenAI deals wouldn’t have had the time to be tested for value at this point. Thus, I don’t think we’ll see a transformation in the reported earnings yet. HPE’s absorbing of Juniper is only starting to see headcount efficiency adjustments, so those won’t yet be reflected, and the sales cycle for network gear is longer than the quarter period, so we won’t likely see too many tangible results. This to me means that there isn’t a huge risk of an AI trough swallowing any of the vendors…yet.
What we should look for, I think, is whether any of the tech companies start to position Wall Street for a coming shift. It’s possible that there will be some impact of the strategic shifts in AI in Q4 of this year, so guidance might be adjusted. If we see much of a shift in guidance now, it likely means that the company doing the shifting is expecting something significant, positive or negative, in Q4.
This sort of expectation might also drive additional announcements of deals or even M&A prior to any earnings report. The thing to look for here is any indication of a recognition of the emphasis shift toward agents, to tools versus models. This would suggest the company doing the deal may be prepping for the new state of the AI market, which would indicate they see what enterprises are seeing and accept the trend as real.
The two companies to look hardest at, at the high level of giant AI, are Google and OpenAI. Google’s Gemini is already making a transition from a model-centric to tool-supported form, and Google is tied with OpenAI in terms of the number of enterprises who spontaneously mention them. OpenAI, as noted above, seems to be doing some deals in anticipation of a tool-centric agent shift in AI, and more comment on that or M&A indicative of it, would be important. They’re still private, so there’s no formal need to report as there will be for Google/Alphabet.
On the network side, Cisco and HPE will be announcing in November and December, respectively, and for them the thing that will bear watching is guidance on AI-centric data center networking. If they are shy about the future there, it will be an indication that the generative-giant players are likely starting to plateau in spending, before we could really expect enterprises to drive up agent deployments.
It’s also important to watch the demand side of the tech world, and in particular the mass-market players like Amazon and Walmart. This sector has been pioneering AI agents, digital twins, and the symbiosis between the two. I’ve heard a lot about this movement in the retail space, and I think it’s enough of a deal to expect it to be reflected in earnings calls. What’s interesting about the retail vertical is the fact that there’s a fusion of transportation, warehousing, retail shelf management and inventory optimization, and sometimes even manufacturing. This all cries out for real-time management and AI agent utilization, as some firms have already told me privately. I contacted Walmart’s PR people to ask for interviews on their use of digital twins and AI agents, but after I submitted a list of questions they changed their minds about helping. Could it be that they don’t want too much revealed to other retailers? Watch the announcements from this vertical carefully.
Don’t neglect Uber, a company whose profit might well be more dependent on effective use of AI than any other company. What they say about their experience, and any comments they make regarding specific AI partners, could be important as an assessment of whether enterprise AI could take off in the transportation vertical.
The chip sector bears watching. Nvidia is a company that people either think is about to hit ten-trillion market cap or about to massively disappoint, but they’re not the only chip player. I think they’ll still show the kind of growth the Street expects in Q3, but their guidance will be important because they’re more likely to play in the giant-AI data center space. AMD might be a better indicator of what’s happening in the agent space, since agents today are likely to be self-hosted by enterprises, and distributed more to point-of-process, demanding more cost-efficient chips. Broadcom has become a bellwether for the AI network space because of its advanced switching chips. A sign of agent deployment might come from more Ethernet switching chips from Broadcom and less Fiber Channel from NVIDIA, and of course from positive comments on data center switching from Cisco, HPE, and other network vendors. Now, OpenAI has done its first custom chip-building deal, with Broadcom, and OpenAI will also integrate Broadcom’s networking chips into racks.
In the networking space, Extreme Networks might be an interesting company to watch, because they depend more than the giants on self-hosted AI deployments of the type agents would represent. The question is whether their likely target companies would be in the forefront of an agent deployment; I don’t have enough data to answer that.
There’s a lot more chances of real movement showing up in Q4 than in Q3, but nobody wants to wait for the shoe to drop before starting to get out of the way, or into it. This may be the alerting quarter, so listen up and get ready.
