JPMorgan revamps AI ‘stocks to buy’ list ahead of earnings | Official Website of Louis Velazquez, entrepreneur, finance guy and tech innovator

JPMorgan just shook up its AI stocks-to-buy list heading into earnings season, positioning Nvidia (NVDA), Broadcom (AVGO), and Micron (MU) at the top of the heap.

The big bank said expectations for the upcoming quarter and beyond remain mostly constructive, despite ballooning valuations across the AI trade.

The note comes at a point when stock market investors will be nursing their wounds from Tuesday’s sell-off, Reuters reported, which sent the major indices sharply lower.

S&P 500: Dropped 2.06% to 6,796.86
Nasdaq Composite: Slid 2.39% to 22,954.32
Dow Jones Industrial Average: Fell 1.76% to 48,488.59

For perspective, with the S&P 500’s total market cap around $58.44 trillion, a 2% drop implies nearly $1.2 trillion in value is erased in just one session.

Markets steadied on Wednesday, Jan. 21, though, with Reuters reporting that the S&P 500 rose 1.16%, the Dow gained 1.21%, and the Nasdaq climbed 1.18%, as tariff jitters cooled.

However, the move wasn’t enough to undo Jan. 20’s damage. Despite a $0.68 trillion recovery in market cap, more than $0.5 trillion is still missing.

The relief bounce does show the AI trade is still in place, although it’s clearly entering a more selective phase.

As I discussed in an earlier story, JPMorgan feels that years of outsized gains are being replaced by a more selective phase.

That makes its revamped list of AI stocks to buy all the more pertinent as we head into a pivotal earnings season with stakes running high.

JPMorgan refreshes AI stock picks as Nvidia, Broadcom, and Micron headline earnings-season focus for investors.

Photo by Woohae Cho on Getty Images

JPMorgan’s AI chip “stocks to buy” list

Nvidia, Micron, and Broadcom lead the list, with JPMorgan leaning toward core AI infrastructure names and a few small-cap standouts.

Nvidia: Anchors the AI trade, led by data center dominance and unrelenting hyperscaler demand
Micron Technology: The big beneficiary of the AI-driven memory demand/supply imbalance
Broadcom: Custom chips and networking exposure positions it squarely at the center of AI buildouts
Marvell Technology: Linked to AI networking and tailor-made compute growth
Analog Devices: Tied to AI through power management, along with industrial demand
KLA Corp: A picks-and-shovels play on advanced chip manufacturing
Synopsys: Mission-critical design software with AI chips is becoming increasingly complex

Small-cap AI chip names JPMorgan also likes:

MACOM Technology Solutions: A networking and RF play linked with AI infrastructure
Astera Labs: A newer AI connectivity that’s currently riding the data center expansion wave

JP Morgan thinks the chip rally still has legs

JPMorgan believes the semiconductor rally still has legs, even as expectations are sky-high heading into earnings.

In a note to clients, analysts led by Harlan Sur laid out a power-packed scenario in which robust execution, solid demand, and constructive guidance will keep the momentum alive this year.

Digging deeper, the bank sees “considerable upside” momentum in the AI accelerator space, led by an eye-popping 50% compound annual growth rate, backed by $200 billion+ in spending in 2025.

However, the optimism comes with some caveats.

Though demand remains healthy and inventories remain lean, JPMorgan warns that the relentless increase in memory prices will continue to pressure PC and smartphone demand later in the year.

I found that out firsthand the hard way, after a recent purchase of a PS5 memory upgrade.

Nvidia’s AI edge is still “the platform”

Next earnings: Feb. 25, 2026 (fiscal Q4 FY2026)

Nvidia is clearly a no-brainer pick in any AI stocks-to-buy list.

Though its stock has been under duress from barbs by the Michael Burrys of the world and other Wall Street pundits, the tailwinds driving demand for its AI stack are unprecedented.

Nvidia currently dominates 80% to 90% of the AI accelerator market, thanks to CUDA, networking, and systems software, which have effectively locked in its customers in a remarkably sticky ecosystem.

More Nvidia:

In its Q3 report, Nvidia posted record sales of $57 billion and Data Center revenue of $51.2 billion, backed by EPS of $1.30.

At the time, Nvidia still finds itself in the thick of the Blackwell ramp, where a confluence of factors including supply, lead times, and mix can effectively swing results.

At the same time, its next platform, Vera Rubin (a rack-scale system), has been announced and is billed as slashing inference token costs dramatically compared to Blackwell and improving training efficiency for AI models.

Broadcom is the custom-AI and networking “quiet compounder”

Next earnings: March 4, 2026, after market close (fiscal Q1 FY2026)

Broadcom isn’t an Nvidia, making headlines for its flashy chips.

Instead, it’s involved in the plumbing that powers massive AI systems. The tech giant builds customized chips for hyperscalers while providing networking gear connecting GPU clusters to operate as a single system.

In the custom chip space, it’s considered a bellwether, with estimates suggesting it dominates 55% to 60% of that niche.

Its latest results were a blowout, where it posted record sales of $18 billion (up 28% year over year).

CEO Hock Tan underscored the outsized impact of AI chip sales (up 74% year over year), pushing overall revenues to record highs. It also told investors it expects AI momentum to continue, projecting AI semiconductor sales to reach $8.2 billion in Q1 and guiding Q1 revenue to be at $19.1 billion.

As we look ahead, Broadcom will continue threading the needle, effectively scaling AI chips while defending margins.

Micron is the “picks-and-shovels” AI memory story

Next earnings: expected around March 19, 2026 (fiscal Q2 FY2026)

Micron has clearly been on every investor’s mind, and why not? The stock has returned a superb 77% in the past three months alone.

Micron’s stock momentum is being driven by a basic supply-and-demand squeeze due to AI. The sprawling data centers powering AI are only getting bigger and faster, and each one needs much more memory than before. That has effectively resulted in an upward reset in customer memory requirements.

The shortage is so severe that, in the medium term, Micron says it can fulfill only about half to two-thirds of demand from some key customers.

CEO Sanjay Mehrotra believes that tightness isn’t something that can be fixed overnight.

Case in point: Micron’s most recent quarter saw it posting $13.64 billion in sales and non-GAAP EPS of $4.78, backed by record free cash flow. As we go deeper into the year, it’s clear that AI memory requirements and the pricing cycle will remain hot, adding to Micron’s operating leverage.