Wall Street rarely changes its mind twice on the same stock within three weeks. With Arista Networks (ANET), it just did.
On May 15, a bullish call flipped Arista from a stock investors had been quietly trimming over customer concentration fears into one of the cleaner AI infrastructure plays heading into 2027.
Yet ANET shares actually fell 3.65% on the day the upgrade was announced, closing at roughly $136.43, according to ChartMill data. That gap, between what the analyst is saying and what the stock price is doing, is the real story.
Here is what changed, who said it, and why the dip may matter more than the upgrade.
Raymond James flips Arista to Outperform on broadening AI demand
The bullish call came from Simon Leopold at Raymond James, who upgraded Arista from Market Perform to Outperform with a $164 price target on May 15, 2026.
Leopold’s thesis rests on three pillars: AI back-end market share gains, an emerging “scale-across” opportunity, and the prediction that Oracle (ORCL) will become Arista’s next 10%-plus customer.
Scale-across, in plain terms, is the networking layer that lets AI training clusters extend across multiple data centers over a Wide Area Network. It is what makes a single training run feel local even when the GPUs are in different cities.
Leopold puts that opportunity at roughly $1 billion in current Arista AI sales, and sees it doubling to $2 billion by 2027.
More AI infrastructure coverage:
- Bank of America tweaks CoreWeave stock price target for 2026
- Broadcom guidance reveals one signal getting unusual attention
- Cloudflare drops eye-opening AI demand numbers after strong quarter
Roughly 40% of Arista’s sales now come from AI applications, with cloud customers contributing another 40%, according to Raymond James. Meta and Microsoft together still represent more than 40% of total revenue.
That is the concentration problem Wall Street has worried about for two years.
Why two upgrades in two weeks signal a shift in Arista’s narrative
This is the second major firm to re-up Arista in a short window. On May 1, Morgan Stanley analyst Meta Marshall raised her price target to $180 from $165 while keeping an Overweight rating, citing the same broadening AI customer base.
Two weeks. Two different firms. Same conclusion.
What changed is the buy-side framing. The story moved from “ANET depends too much on Meta and Microsoft” to “ANET is the picks-and-shovels play on the next AI capex wave.“
If Oracle and Google materialize as 10%-plus customers, the concentration discount Wall Street has been applying to the stock starts to look stale.
That is the bet embedded in both notes.
The Broadcom chip bottleneck behind Arista’s dip on upgrade day
Here is where the story gets uncomfortable for short-term holders.
Leopold himself flagged that Broadcom (AVGO) chip availability may constrain Arista’s shipments through 2026, even as demand keeps building.
Arista’s switches rely heavily on Broadcom’s Tomahawk and Jericho silicon. When Broadcom cannot ship enough chips, Arista cannot ship enough boxes. Revenue gets pushed right, not lost.
That is the “de-commits are delays, not cancellations” language Raymond James used in the note.
Related: Morgan Stanley resets Arista stock price target for 2026
Management on the Q1 2026 earnings call confirmed the same dynamic, citing shortages in semiconductor wafers, memory chips, and CPUs as headwinds for gross margin through the year.
The market sold the upgrade because investors connected the same dots Leopold did: better 2027, choppier 2026.
How Arista’s recent numbers stack up against the broader market
Despite the AI narrative, ANET has been a relative laggard year-to-date.
A quick performance scoreboard versus the S&P 500, reports Barchart data:
ANET vs S&P 500 performance:
- YTD 2026: ANET +4.1% vs S&P 500 +8.3%
- Trailing 12 months: ANET +57.7% vs S&P 500 +31%
- 52-week range:$83.86 to $179.80
So the 12-month chart still looks like an AI winner. The year-to-date chart looks like a stock taking a breather after a parabolic 2025.
That is roughly where the Raymond James $164 target lands: about 20% above the May 15 close, but well below the $220 high target at Morgan Stanley and the $186 analyst consensus across 41 firms.
What needs to happen for the Arista bull case to play out
The setup is increasingly clean, but it is not automatic. A few specific milestones need to land for Arista’s 2027 thesis to translate into a re-rated stock.
Four catalysts that would validate the upgrade:
- Oracle crosses the 10% revenue threshold, confirming the customer diversification story
- Broadcom relieves supply constraints in 2027, unlocking the shipment backlog
- Scale-across revenue doubles to $2 billion, as Raymond James models
- Q2 2026 guidance holds at $2.8 billion, the Arista-issued framework for the current quarter
The first two are largely outside Arista’s control. The last two are not.
Arista already raised its full-year 2026 revenue outlook to $11.5 billion, implying roughly 27.7% growth. AI networking revenue alone is now guided to $3.5 billion for the year, up from a prior $3.25 billion target.
If those numbers hold through Q2 and Q3, the supply-driven dip starts to look like a buyable setup rather than a structural problem.
How retail investors can think about the Arista setup
The Raymond James note effectively splits Arista into two trades.
For investors with a 12- to 18-month horizon, the chip bottleneck is a feature, not a bug. It compresses near-term shipments into a 2027 capacity expansion, which is exactly when scale-across revenue is set to double.
Related: Jim Cramer spots hidden upside in a surging chip stock
For investors trading the next earnings cycle, the same bottleneck is a real margin and visibility risk. Gross margins are guided in the 62% to 63% range, with potential pressure from memory and silicon costs.
A few practical things to monitor before adding exposure:
- Insider activity:170 insider sales versus zero purchases over the past six months is a meaningful caution flag
- Nvidia‘s Ethernet entry, which is a real competitive overhang on Arista’s switching dominance
- Broadcom’s own supply commentary on its next earnings call, since AVGO chip availability is the single biggest variable in Arista’s 2026 revenue line
None of these break the long-term thesis. They just shape the entry point.
The takeaway for Arista stock investors
Two Wall Street upgrades in two weeks rarely happen for a stock about which the market is openly skeptical.
The fact that Raymond James and Morgan Stanley are now aligned on the customer diversification story matters,aa because it changes the valuation multiple investors are willing to pay for the stock.
What also matters: the analyst who raised the target is the same analyst flagging the supply problem. Leopold is not saying buy ANET today and ride it. He is saying the 2027 setup is strong enough to look through the 2026 noise.
That is a different argument than a typical “buy on strength” upgrade, and it is why the stock fell on the news.
For long-term investors, the dip on upgrade day may be the more useful signal than the upgrade itself.
For traders, the next inflection point is Arista’s Q2 print in early August, where guidance on Broadcom supply and Oracle’s revenue trajectory will either confirm the new bull case or send the stock back into its concentration-risk box.
