Is Cisco a Buy Before Wednesday's Earnings? Not at $121, Not Before the AI Guide

Updated August 12, after the close: both triggers hit, and the price reset too

Both of the specific things that would have made me a buyer showed up. Q4 AI orders came in at $4.0 billion, above the roughly $3.7 billion this piece said the raised full-year target required, and the FY27 AI revenue guide landed at $7.5 billion, well above the $6 billion floor. Full numbers are in the updated earnings preview.

What I did not expect: the stock fell anyway. Shares closed the regular session near $123.88, then slipped roughly 2% after hours once the numbers were out, a sell-the-news reaction on a report with no weak line in it. That is an early after-hours read, not a settled close, and it can move again before Thursday's open.

If that level holds anywhere close to where it printed, roughly the low-$120s, this is the pullback I said I wanted: a confirmed FY27 number instead of a guess, at a price close to where the stock sat a week ago rather than the peak it built into the print. I would rather own it here than at $121 the day before the guide, because the guide is no longer a coin flip. I'm not calling the exact after-hours print the entry; I'm saying the setup this piece asked for, a reset price with the AI number confirmed, is close to what showed up. Worth rechecking against Thursday's actual open before acting on it.

TL;DR

• Not before Wednesday. The stock is $121.43 after a roughly 46% run this year, the quarter is fenced by management's own guide, and the entire remaining bet is a first-ever formal fiscal 2027 AI revenue guide that nobody has seen.
• The re-rating is not fake: AI infrastructure orders tripled year over year in Q3 and networking revenue grew 25%. The question is not whether Cisco's AI business is real. It is whether $121 already pays for it.
• The entry I would rather own is after the guide, even at a slightly higher price: a confirmed number converts the story into arithmetic, and paying up for certainty beats paying less for a coin flip on a stock at cycle-high multiples of its own history.

Is Cisco a Buy Before Earnings?

No. Not because the business is in trouble, but because the reward for guessing right on Wednesday is small and the punishment for an "in line" guide is not. The full preview lays out the mechanics: consensus of $1.17 on $16.83 billion sits exactly on the midpoint of management's guide, so the print cannot surprise much in either direction. Everything rides on the fiscal 2027 guidance, where management has promised to formalise at least $6 billion of hyperscaler AI revenue.

A stock up 46% in a year has already extended a lot of credit to that number. If the guide lands at the $6 billion floor, "as promised" can easily trade as "priced in," and the day goes red on a perfectly good report. That is the asymmetry: modest upside for a strong guide the market half-expects, real downside for an adequate one.

What Would Make Me a Buyer

Specific things, all checkable Wednesday evening:

• Q4 AI orders near $3.7 billion, which is what the raised $9 billion full-year target arithmetically requires. That would mean the order book accelerated again into year-end.
• An FY27 AI revenue guide meaningfully above the $6 billion floor, because the floor is already in the price.
• A pullback that resets the entry. The stock is about 6% below its June record. A guide-day wobble to the low $110s with the AI numbers intact is the setup where the re-rating and the price agree again.

The bear case beyond the price: the restructuring (about $450 million of charges landing this quarter) makes GAAP numbers noisy for two more quarters, and Cisco's AI revenue remains a fraction of the networking base, so the growth story is still carried by a minority of the business. Neither kills the thesis. Both argue against paying peak-credit prices the day before the evidence.

The One-Line Read

Cisco at $121.43 is a good business at a price that assumes Wednesday goes well, and I would rather pay $125 for a confirmed fiscal 2027 AI number than $121 for an unconfirmed one: the trade is to let the guide print first, because after a 46% year the market has stopped paying for promises and started grading them.