Nvidia, Intel, Micron, MPS Named Buys Ahead of Q3 Earnings [2026]

KeyBanc Capital Markets analyst John Vinh named Nvidia, Intel, Micron Technology, and Monolithic Power Systems as his top semiconductor stock picks heading into the third quarter 2026 earnings cycle, according to a note covered by Barron’s and Markets Insider. The call lands at an unusual moment for chip investors, with AI infrastructure spending still accelerating, memory prices climbing on constrained supply, and all four companies set to report results within a four week window of each other. For developers, engineering leaders, and anyone building on top of AI coding tools, this earnings stretch is less a Wall Street curiosity than a preview of what GPU access, memory costs, and compute pricing will look like heading into 2027.

The timing matters. Monolithic Power Systems and Intel are both scheduled to report on October 29, 2026, while Nvidia’s fiscal third quarter 2027 results land on November 17, 2026. Micron already reported earlier in its own fiscal calendar, and its numbers are a big part of why Vinh’s thesis has teeth: a 29.8% quarter over quarter jump in memory unit shipments paired with a 24.1% price increase. That combination, tight supply and rising prices at the same time, is exactly the setup chip analysts watch for when they want to call a cycle rather than just a quarter.

·

Google · Preferred Sources

Don't miss new tech stories on Google

Add TrendinTech once in the Google app and our stories appear in your news suggestions.

Add Now

Who KeyBanc Picked and Why It Matters Now

Vinh’s list is notable for what it includes beyond the obvious AI darling. Nvidia’s presence on a buy list ahead of earnings is not surprising given its dominance in AI accelerators. What stands out is the inclusion of Intel, a company that has spent much of the past three years working through a turnaround, and Monolithic Power Systems, a power management chipmaker that rarely gets mainstream coverage outside of semiconductor trade press. Micron rounds out the group as the memory play, benefiting from what the report describes as constrained capacity across DRAM, high bandwidth memory, and NAND simultaneously.

That combination tells a story about where the next leg of AI infrastructure spending is actually going. It is not just about who makes the fastest GPU anymore. It is about who controls the memory bandwidth those GPUs need, who supplies the CPUs that orchestrate workloads around them, and who builds the power delivery systems that keep increasingly dense server racks from melting down. Readers who followed the Micron HBM4 memory roadmap webinar will recognize this pattern: the AI compute story has quietly become a memory and power story too.

Wall Street Estimates Going Into the Reports

The consensus numbers heading into the reporting window give a sense of scale. Analysts are modeling Monolithic Power Systems to post earnings per share of $7.71 on revenue of $1.15 billion. Intel’s third quarter estimates sit at roughly $0.40 in earnings per share against $16.48 billion in revenue. Nvidia, reporting its fiscal third quarter 2027 numbers in mid November, carries the highest bar by far: $2.47 in earnings per share on close to $108.80 billion in revenue.

Company	Ticker	Report Date	EPS Estimate	Revenue Estimate
Nvidia	NVDA	November 17, 2026	$2.47	~$108.80 billion
Intel	INTC	October 29, 2026	~$0.40	~$16.48 billion
Monolithic Power Systems	MPWR	October 29, 2026	$7.71	~$1.15 billion
Micron Technology	MU	Reported prior quarter	Shipments +29.8% QoQ	Prices +24.1% QoQ

That revenue figure for Nvidia is worth sitting with for a second. A single company is expected to report close to $109 billion in quarterly revenue, a number that would have seemed absurd for any chipmaker just a few years ago. It is also a figure that developers building AI powered products should pay attention to, because Nvidia’s pricing power and allocation decisions trickle directly into what cloud providers charge for GPU instances, which in turn shapes the economics of every AI coding assistant and inference heavy developer tool on the market.

Micron’s Memory Squeeze and What It Means for AI Infrastructure

Micron’s recent quarter is arguably the clearest signal in this entire story. Memory unit shipments rose 29.8% quarter over quarter while prices climbed 24.1% in the same period. Rising shipments and rising prices moving together usually means demand is outrunning supply, not that the company is simply selling more at the same price point. Micron CEO Sanjay Mehrotra has pointed to the company’s roadmap for future HBM products, describing work with Nvidia on what he called the industry’s first custom HBM implementation.

That custom HBM collaboration is a meaningful detail for anyone tracking the AI hardware stack. High bandwidth memory has become the bottleneck component in modern AI accelerators, more so in many cases than raw compute. If Micron and Nvidia are co-designing memory specifically for upcoming accelerator generations, that is a sign the two companies expect demand for AI training and inference capacity to keep climbing well past the current earnings cycle. It also explains why Micron shows up on a buy list even though it is not a household name among the AI companies most developers interact with directly.

The memory squeeze has a direct line to developer tooling costs. Every large language model inference call, whether it is powering a chatbot or an AI coding agent reviewing a pull request, depends on memory bandwidth to move model weights and activations fast enough to keep latency acceptable. When HBM supply tightens, cloud providers pay more to build out capacity, and that cost eventually shows up somewhere in the pricing of AI developer tools, directly or through tighter rate limits and allocation caps.

Intel’s Turnaround Bet: Server CPUs in an AI World

Intel’s inclusion on this list is the most debated of the four, and for good reason. The company has spent several years rebuilding its foundry ambitions and fighting to hold server CPU market share against AMD, all while watching Nvidia capture the lion’s share of AI infrastructure spending growth. Intel’s October 29 earnings report, landing the same day as Monolithic Power Systems, will be read closely for signs of whether its server and data center business is stabilizing or still shrinking.

What makes Intel relevant to a coding tools and developer audience specifically is less about GPUs and more about the orchestration layer. Every AI data center still runs on CPUs for networking, storage, preprocessing, and general workload management around the GPU clusters doing the heavy lifting. A company that previously covered Nvidia’s shrinking China AI chip market share against Huawei noted how competitive pressure in one region can reshape a chipmaker’s global strategy fast. Intel faces a similar dynamic domestically: it needs server wins in AI adjacent infrastructure to prove its turnaround story has legs beyond cost cutting.

Monolithic Power Systems: The Quiet Power Play

Monolithic Power Systems is the least recognizable name on Vinh’s list to anyone outside semiconductor investing circles, but its business is arguably the most structurally tied to the AI buildout’s physical constraints. The company makes power management chips, the components that convert and distribute electricity inside servers and data centers. As AI racks get denser and draw more power per square foot, the demand for efficient power delivery components grows in lockstep, independent of which GPU vendor wins any given generation.

Monolithic Power’s October 29 earnings date puts it on the same day as Intel, meaning the market will get two very different reads on AI infrastructure health within hours of each other: one from a legacy CPU giant trying to prove relevance, and one from a specialized power components supplier riding the AI data center wave almost by default. Wall Street’s $7.71 earnings per share estimate against $1.15 billion in revenue suggests expectations are already baked in for continued strength.

Why This Earnings Cycle Matters for AI Coding Tools

It is worth connecting the dots explicitly for readers who build or use AI coding tools rather than trade semiconductor stocks. The explosion of AI coding assistants over the past two years, from the kind of tooling covered in our look at GitHub Copilot and Cursor’s expanding AI coding race to agent platforms like those discussed in our OpenCode 2.0 API access coverage, runs entirely on the infrastructure these four companies supply. Every autocomplete suggestion, every AI generated pull request review like the kind featured in our piece on Claude Code’s recent update cycle, consumes GPU cycles, memory bandwidth, and electricity that trace back to Nvidia, Micron, Intel, and companies like Monolithic Power Systems.

When memory gets more expensive and GPU allocation gets tighter, the companies building AI coding tools face a choice: absorb the cost, pass it on through higher subscription prices, or ration access through usage limits. Readers tracking tools built on cheaper inference, such as the pricing comparison in our Mistral Large 4 versus GPT-6 Astra benchmark, have already seen how aggressively vendors are competing on price even as underlying hardware costs climb. That tension, cheaper AI products built on more expensive infrastructure, is the real story underneath Vinh’s buy list.

Historical Context: How We Got Here

The semiconductor industry has gone through several distinct cycles over the past decade, but the current one is unusual in how concentrated the growth has become. Gartner’s worldwide semiconductor revenue report found that total industry revenue hit $793 billion in 2025, up 21% year over year, a figure that itself reflects how much of the broader technology economy now runs through chip spending. Within that total, Nvidia generated $125.703 billion in semiconductor revenue in 2025, representing 15.8% of the worldwide semiconductor market and growth of 63.9% year over year, according to Gartner. That single company’s share of the entire global semiconductor market jumping to nearly one in six dollars spent is a scale of concentration the industry has not seen from a single vendor in modern memory.

Intel’s 2025 numbers tell the opposite story: Gartner recorded $47.883 billion in semiconductor revenue, good for 6.0% worldwide market share, but a decline of 3.9% year over year. That contraction, even as the overall market grew by a fifth, is the backdrop against which Intel’s turnaround efforts and its inclusion on Vinh’s buy list need to be read. Micron, meanwhile, posted $41.487 billion in 2025 semiconductor revenue per Gartner, a 5.2% market share figure that grew 50.2% year over year, nearly matching Nvidia’s percentage growth rate off a much smaller base.

Previous chip cycles were driven by smartphones, then by the crypto mining boom, then by pandemic era remote work hardware demand. Each of those cycles eventually cooled as the underlying catalyst matured or collapsed. The current AI driven cycle has already outlasted the typical eighteen to twenty four month window of prior booms, and Vinh’s bet is effectively a wager that it still has room to run through at least another two quarters.

Competitive Comparison: Four Different Bets on the Same Trend

What makes Vinh’s four stock picks interesting as a group is that they are not really competitors with each other. Nvidia dominates AI accelerators, Micron supplies the memory those accelerators depend on, Intel competes in adjacent server CPU and orchestration markets, and Monolithic Power Systems sells power management components to practically everyone in the data center supply chain, including, in some cases, Nvidia’s own server partners. That lack of direct competition is itself part of the thesis: a single buy list spanning the full AI infrastructure stack reduces the risk of any one company’s execution stumble tanking the whole bet.

Company	Primary Role in AI Stack	2025 Semiconductor Revenue	2025 YoY Growth
Nvidia	AI accelerators / GPUs	$125.703 billion	63.9%
Intel	Server CPUs / orchestration	$47.883 billion	-3.9%
Micron Technology	DRAM, HBM, NAND memory	$41.487 billion	50.2%
Monolithic Power Systems	Power management components	Not separately reported by Gartner	Not separately reported by Gartner

That table also makes clear why Intel’s inclusion draws the most skepticism among analysts: it is the only one of the three companies with separately reported Gartner figures that actually shrank in 2025 while the rest of the market grew by a fifth. Vinh’s buy case rests on the idea that Intel’s server business has stabilized enough to turn that decline around, a bet that will get its first real test on October 29.

What Analysts Are Watching For

Beyond the headline revenue and earnings per share numbers, there are a handful of specific metrics that will determine whether Vinh’s thesis holds up. For Micron, the question is whether the 29.8% quarter over quarter shipment growth and 24.1% price increase from the prior quarter can continue, or whether supply catches up and prices soften. For Intel, investors will be watching server CPU unit volumes and any commentary on foundry customer wins, since that business has been the linchpin of its multi year turnaround narrative. For Monolithic Power Systems, the key signal is whether data center power management revenue keeps outpacing the company’s more traditional industrial and automotive segments.

For Nvidia, given its scale, the market will be parsing data center segment revenue specifically, along with any guidance on GPU allocation and supply constraints heading into 2027. A beat on the headline $108.80 billion revenue estimate would reinforce the AI infrastructure growth story broadly; a miss, even a small one, could trigger outsized volatility given how much of the broader chip sector’s momentum has been riding on Nvidia’s results as a bellwether.

Risks to the Buy Thesis

No buy call is without risk, and this one carries a few specific to each name. Micron’s memory cycle has historically been volatile, with pricing swings that can reverse quickly if customers over order during a shortage and then work through excess inventory. Intel’s turnaround remains unproven at the earnings level; a Hold rating from the broader analyst consensus on the stock reflects real skepticism that has not gone away just because it made KeyBanc’s buy list. Monolithic Power Systems, despite its data center tailwind, is a smaller company whose results can be more volatile on a percentage basis than its larger peers. Nvidia carries perhaps the least company specific risk among the four, but the highest valuation sensitivity: expectations are now so elevated that even a solid quarter can disappoint if guidance for the following period comes in merely in line rather than ahead.

Five Predictions for the Earnings Window

• Micron’s memory pricing momentum will carry into its next reported quarter, with HBM revenue growth outpacing the company’s legacy DRAM and NAND segments as the Nvidia custom HBM collaboration scales.
• Intel’s October 29 report will show incremental stabilization in its server CPU business rather than a dramatic turnaround, keeping its consensus rating a point of debate among analysts rather than resolving it.
• Monolithic Power Systems will post another quarter of data center revenue growth outpacing its industrial and automotive segments, reinforcing its role as a power delivery proxy for AI infrastructure buildout.
• Nvidia’s November 17 report will draw outsized market attention regardless of outcome, given its position as the most closely watched single data point for AI infrastructure spending sentiment heading into 2027.
• AI coding tool and developer platform pricing will continue to diverge from underlying hardware costs in the near term, as vendors compete on price even as the memory and GPU supply chain underneath them tightens.

What Developers and Engineering Leaders Should Watch

For engineering teams budgeting for AI tooling in 2027, this earnings cycle is a useful leading indicator. If Micron’s memory pricing trend continues and Nvidia’s data center revenue keeps climbing at anywhere close to its 2025 growth rate, the cost of running inference heavy workloads, including the AI coding agents and review tools increasingly embedded in software development pipelines, is unlikely to get cheaper purely from a hardware supply standpoint. Any price relief developers see will come from software efficiency gains and vendor competition, not from looser chip supply. Teams that rely on GPU accelerated inference at scale, whether for internal AI coding assistants or customer facing products, should treat this earnings window as a signal to lock in capacity or pricing where they can rather than assume costs will drift lower on their own.

Cloud infrastructure teams that read our earlier coverage of Nvidia shifting its GB202 silicon away from consumer RTX 5090 cards toward RTX Pro and data center allocations already have a preview of how tight the company’s prioritization decisions can get when enterprise demand competes with consumer product lines. That same dynamic, prioritizing the highest margin, highest demand customer segment, is likely to repeat across this earnings cycle as all four companies report.

Frequently Asked Questions

Who named Nvidia, Intel, Micron, and Monolithic Power Systems as buy picks?
KeyBanc Capital Markets analyst John Vinh named the four companies as his top semiconductor stock picks ahead of third quarter 2026 earnings, according to reporting from Barron’s and Markets Insider.

When do these companies report third quarter 2026 earnings?
Monolithic Power Systems and Intel are both scheduled to report on October 29, 2026. Nvidia reports its fiscal third quarter 2027 results on November 17, 2026.

What are the Wall Street earnings estimates for these companies?
Estimates cited include $7.71 earnings per share and $1.15 billion in revenue for Monolithic Power Systems, roughly $0.40 earnings per share and $16.48 billion in revenue for Intel, and $2.47 earnings per share with approximately $108.80 billion in revenue for Nvidia.

Why is Micron included despite not reporting in the same window?
Micron’s prior quarter showed memory unit shipments rising 29.8% quarter over quarter alongside a 24.1% price increase, a combination analysts read as evidence of tight DRAM, HBM, and NAND capacity supporting continued pricing power.

What did Micron’s CEO say about working with Nvidia?
Micron CEO Sanjay Mehrotra said the company has a strong roadmap for future HBM products and is working with Nvidia on what he described as the industry’s first custom HBM implementation.

How does this earnings cycle affect AI coding tools?
AI coding assistants and developer agent platforms run on GPU and memory infrastructure supplied largely by these companies. Tighter memory supply and rising GPU allocation pressure can influence the pricing and usage limits that AI coding tool vendors set for developers.

Is Intel considered as strong a buy as the other three companies?
Intel’s inclusion on the list has drawn more debate than the other three names, reflecting ongoing questions about its server CPU turnaround relative to Nvidia, Micron, and Monolithic Power Systems, which have each posted stronger recent growth.

What role does Monolithic Power Systems play in the AI buildout?
Monolithic Power Systems supplies power management components used to convert and distribute electricity inside AI servers and data centers, a role that grows in importance as AI server racks draw more power per unit of compute.

Related Coverage

• Samsung Taps TSMC 3nm Lines for Nvidia Next HBM [2026]
• NVIDIA RTX Spark Laptop Price: $2,399 vs $2,599 [2026]
• NVIDIA RTX Spark PCs Hit $2,599, 6 Makers Ship Day 1 [2026]
• NVIDIA Run:ai Goes Open Source With 4 GPU Tools [2026]
• Gigabyte BIOS Hints at Intel 2027 CPUs With Dual RAM [2026]