JPMorgan is staying Overweight with a $580 target — nearly 50% above where the stock sits after round-tripping a 17% post-earnings drawdown. The market priced in a broken story; Hock Tan's numbers say otherwise: AI semiconductor revenue up 143% year over year, Q3 guided to $16B, and backlog visibility now stretches into 2028.

The single hottest chart on this list. Evercore initiated at Outperform with a $325 target last week — sitting above an already-bullish $272 street average — and Stifel, BofA, and Goldman all hiked targets inside the same seven days. AI-connectivity demand for Credo's copper-and-optical links is outrunning even optimistic models.

JPMorgan keeps MRVL as one of its three AI-chip Overweights alongside NVDA and AVGO, citing the shift to faster AI networking — Marvell's optical business is projected to grow roughly 70% year over year. Cantor and Stifel both raised targets into the $300–$350 range in the last three weeks.

20 of 29 analysts rate this Strong Buy, with an average target 45% above today's price. Optical components for AI data centers — the same trade as CRDO, one layer up the stack.

Strong Buy from 9 of 12 analysts, average target 41% above today's $205.87. This is the highest-conviction “chip alternative to Nvidia” story on the board, and it's still down sharply from its post-IPO high — which is exactly where the risk/reward gets interesting.

The anchor of the list. Foxconn says AI-rack shipments and accelerator demand are staying strong into Q3 — exactly the read-through the whole sector is watching for. Fifty-five of fifty-seven covering analysts rate it Strong Buy, the least controversial demand story on this screen.

Carrying this one forward. The post-earnings crash bottomed near $161 and the stock has ripped back into the mid-$200s — and even after the bounce, the entire analyst range now sits at or above today’s price.

The best-timed name on the board. TSMC reports Q2 earnings July 16 — eleven days out — and just raised full-year 2026 capex guidance to a record $52–56B, with more than 70% earmarked for advanced nodes.

Still the name that started this whole sector’s selloff back in June — and still working through it. Shares slipped again this week even as the Jalapeño custom-chip story with OpenAI keeps developing.

Got caught in this week’s broad chip-sector selloff, giving back some of its post-earnings run. The fundamentals didn’t move — the Y1.5T Hiroshima fab expansion for AI-grade DRAM and HBM is still on track for 2028 shipments.

Fell harder than almost anything else on this list this week — down over 11% with no company-specific bad news, pure sector rotation out of optics and photonics names. The floor still sits well above today’s price.

My highest-conviction, highest-risk call. The Digital Credit Capital Framework from last week — preferred and common buybacks, a bigger USD reserve, a dividend hike — directly targets the NAV-discount problem that’s had this stock trading below its own Bitcoin holdings.

A great business trading like it. AMD is up huge this year and the average analyst target has basically caught up to the stock — near-term upside from here depends on the next catalyst, not the target catching up to the price.

The steadiest mega-cap on this list — the most analyst coverage of anything here, and the least controversial thesis. Cloud growth and the Gemini build-out keep grinding higher without needing a single headline to justify the position.

High-quality, high-conviction, and quiet this week — which is the point. No single headline moved this one; it just kept compounding while noisier AI names whipsawed.

AWS’s Trainium buildout keeps this one in the AI conversation, but the more interesting story this week is the growing local pushback against new data-center construction in several markets.

A quality enterprise-software name with a wide analyst-implied range. This week’s news is modest — a regional automation partnership — but the setup underneath it is clean.

The only name on this list where the analyst floor sits below today’s price — worth sizing accordingly. A board-level governance change this week isn’t a growth catalyst, just noise worth knowing about.

Last week’s steadiest performer — down just 4.56% in a red week — gets a fresh vote of confidence this week.

The widest range on this list — and the most binary. A new Nvidia partnership for sovereign AI deployment plus a fresh U.S. Army Foundry contract reignited the bull case this week, but the daily chart is still technically bearish, sitting below all three major moving averages.

The rare setup where the entire analyst range sits above today’s price — even the most bearish target implies +50%. First earnings as a public company (June 23) showed 92% revenue growth and confirmed a $20B+ multi-year OpenAI deal, but Q2 margin guidance of 36–38% (down from 46.5%) spooked the stock into a 16–17% drop before it bounced ~8% into the weekend.

The single biggest catalyst on the board. Rocket Lab agreed today to acquire Iridium Communications for ~$8.0B in cash and stock, instantly converting the company from a launch provider into a vertically integrated space player with Iridium’s 66 active LEO satellites and 2.5M+ billable subscribers. Shares jumped as much as 15.6% intraday on the news.

Caught in the sector-wide AI selloff with no company-specific bad news — which is exactly what makes it interesting. Lumentum sits at the center of Nvidia’s $6.5B+ photonics buildout, anchored by Nvidia’s own $2B investment in the company. The floor on this one sits above today’s price.

The name that started the whole sector's selloff in early June now has the update that could turn sentiment back around. Broadcom and OpenAI unveiled Jalapeño, OpenAI’s first custom inference chip, co-developed in nine months for gigawatt-scale deployment by end of 2026. CEO Hock Tan called compute demand from customers “simply insatiable.”

The steadiest name on the list, with a real technical headline behind it: IBM unveiled the world’s first sub-1nm chip architecture on June 25, claiming up to 50% more performance or 70% better energy efficiency than its 2nm node. JPMorgan upgraded to Overweight two days earlier, citing software acceleration into H2.