5 Dirt Cheap AI Stocks To Buy In 2026

5 Dirt Cheap AI Stocks To Buy In 2026 The dot com bubble and the AI trade look similar. The difference is that this time the earnings showed up. And when earnings grow fast enough for long enough, something happens to valuations. The dot-com bubble and the AI trade look similar from the outside. Both produced enormous stock moves, both generated relentless headlines, and both left a lot of investors feeling like they’d arrived too late. The difference is that this time the earnings showed up. And when earnings grow fast enough for long enough, something interesting happens to valuations. They compress. A stock that traded at 40x earnings two years ago can trade at 10x today without the price moving much at all, simply because the earnings underneath it have tripled. That’s an expensive stock becoming cheap while nobody was paying attention. The Evening Sun, 1999. We went looking for exactly that. Five AI infrastructure companies whose earnings have grown fast enough, and consistently enough, to compress what were once lofty valuations into some of the cheapest multiples in their respective sectors. In every case a forward P/E that sits well below where the market prices comparable businesses. We think the gap between what these companies are worth and what the market’s currently paying for them won’t stay this wide for a long time. Before we begin, this Substack is for informational and entertainment purposes only and does not constitute financial advice. Past performance is not indicative of future results. Never invest more than you can afford to lose. Our writers may hold positions in the securities discussed and may buy or sell them at any time without notice. 1. Why Forward P/E Is Critical In This Cycle There are many ways to value a stock. Price-to-sales, EV/EBITDA, price-to-book, free cash flow yield The list is long and the debates around each metric are longer. But in a cycle driven by explosive, compounding earnings growth, one metric cuts through the noise faster than anything else: the Forward P/E ratio. The forward P/E is simple. It takes what a company is expected to earn over the next twelve months and divides the current stock price by that number. The result tells you what you’re paying today for tomorrow’s earnings. In a normal earnings environment, the difference between trailing and forward P/E is modest. In an environment where earnings are doubling and tripling YoY, the gap between the two becomes enormous. The investors who anchor to trailing figures end up massively overpaying for businesses that are getting cheaper by the quarter. Source: Yardeni Research That’s the environment we’re in right now. It’s a multi year CapEx cycle that’s running through the income statements of every company supplying the infrastructure. Earnings estimates for the stocks in this article have been revised upward repeatedly over the past 18 months, which means their forward P/Es have been compressing even as their stock prices have moved. The market’s still reading last year’s newspaper. The semiconductor sector trades at 20.2x Fwd earnings today. Technology hardware trades at 26x. Several of the companies we’re covering in this article trade at fractions of those figures, despite growing earnings faster than their peer groups. That’s the setup. Forward P/E is the lens. Everything else follows from it. 2. The Cycle Matters How is it different? Normal cycles have a ramp, a peak, and a digestion period where customers pause, work through their inventory, and wait before ordering again. What’s happening right now looks different. Magnificent 7 have collectively committed hundreds of billions of dollars in CapEx over the next several years, and those commitments are accelerating. Why? The reason is competitive. No hyperscaler can afford to fall behind on compute capacity when their competitors are building at the same pace. The moment one pauses, the others gain ground. That dynamic has turned AI infrastructure spending into something that looks like an arms race, and it means the companies supplying the chips and the servers and the networking equipment are operating in a demand environment that has very little historical precedent. These companies base their growth guidance on firm purchase commitments. Clients and partners have already defined the demand. Long term supply agreements and take-or-pay contracts turn previously cyclical revenue into predictable cash flow. Earnings visibility for several of these names extends well into 2027 and beyond. Infrastructure spending remains at record levels. Massive capital expenditure flows into these businesses every quarter. Yet their forward P/E ratios have compressed harder than ever, and the companies sitting at the center of that compression fall into two distinct parts of the AI supply chain. That’s exactly where we went looking. Sectors Covered We group the companies into two distinct industries: Semiconductors and Semiconductor Equipment. These firms design the chips that process data and the manufacturing gear that builds those chips. They provide the fundamental compute engine for AI workloads. Technology Hardware, Storage, and Peripherals. These firms build the physical infrastructure. They provide the server racks, the storage arrays, and the networking equipment that allow data to move between chips and systems. We’re sure you’re familiar with both. We chose these categories for a specific reason. The market assigns these groups high forward valuation multiples. The semiconductor industry trades at 20.2x. Technology hardware trades at 26x. These numbers reflect strong growth expectations. Yet, we find specific companies within these categories that trade at significant discounts to these already high benchmarks. The following section contains the deep-dive analysis. We break down the financials for five key players and we explain why we believe they are the most mispriced plays in the market. 4. Deep-dive Into Each Stock What follows is a breakdown of 5 companies we believe are the most undervalued plays in AI infrastructure right now. Two industries, five stocks, one shared characteristic: The market’s pricing them like the old cycle when they’re operating in a completely new one. All individual stock forward P/E ratios are sourced from Koyfin. Sector median forward P/E benchmarks are sourced from Yardeni Research, as of June 25, 2026: semiconductors at 20.2x and technology hardware at 26x. You’ll see the full picture in “The Gap” at the end. The rest of this report is available to paid subscribers only . Join hundreds of investors getting access to: Multiple deep dive articles weekly VIP discord server with private chat and daily trade alerts/analysis from 4 analysts Around the clock support/guidance from Michael, Kawz, and AB analyst team Get Michael and Kawz’s $70M Autopilot fund copy trade updates ($400/year separate value included) Read more

Macro/FOMC preview framing: markets pricing an FOMC hold; author argues the prior “capex/hyperscaler AI buildout” support for equities has deteriorated due to higher oil/inflation, persistently high rates, widening credit spreads, and Chinese open-source AI progress compressing margins—creating negative tech sentiment into the meeting. No explicit tickers/cashtags in the post; implications are broad risk-on tech vs energy/rates/credit.

Post argues Qualcomm ($QCOM) has a newly confirmed hyperscaler custom-silicon engagement for data-center CPU with initial shipments later this calendar year, potentially driving an AI/data-center re-rating. It frames $QCOM as a “cheap legacy smartphone chipmaker” (low forward P/E cited) with hidden AI upside, while acknowledging handset demand/memory-shortage risks and secular mobile concerns. Mentions valuation comps ($ARM, $INTC, $AMD) and an analogy to Soitec (Soitec) as prior “hidden AI upside” re-rating example.

Post argues AAOI’s ~47% drawdown (from ~$220 to ~$115 since May 13) is sentiment-driven rather than fundamentals-driven, citing record revenues, higher gross profit dollars, large cash balance, continuing hyperscale orders, and an Amazon warrant tied to cumulative purchases. It attributes the selloff to dilution/ATM filing, insider selling headlines, and an EPS miss amid “priced for perfection” momentum. It also references sector-wide weakness allegedly catalyzed by a SemiAnalysis piece on CPO delays/optical valuations, later refuted by NVIDIA/optical companies, and notes SemiAnalysis/Tema launched a “LAZR photonics ETF” with AAOI as a top holding (implying potential narrative/flow effects).

Post argues the key bottleneck in AI optical interconnect buildout is not lasers/transceivers themselves but the required testing/qualification of every optical component before deployment. It claims a single failed optic can cause large-scale AI cluster downtime costs, implying rising demand/pricing power for optical test & measurement providers. Mentions Nvidia GTC anecdote (CoreWeave CTO complaining about failed optics) and cites massive hyperscaler capex as demand driver.

Post is an earnings-preview style note focused on Applied Optoelectronics ($AAOI) ahead of an imminent earnings report. It frames AAOI as a key beneficiary of a “photonics supercycle,” cites alleged hyperscaler orders and capacity expansion, and mentions a read-through to $LITE. Actionability is moderate: there is a clear near-term catalyst (ER tomorrow) and explicit ticker focus, but much of the post is promotional and performance/positioning talk rather than concrete, checkable forecasts.

Post argues a U.S. federal $2.013B CHIPS Act quantum investment (minority equity stakes across nine quantum companies) is a major catalyst that drives a sector-wide re-rating. It highlights Infleqtion (ticker given as INFQ) as a newly SPAC’d neutral-atom quantum company with government customers, and notes sharp post-announcement price moves across quantum names (INFQ, QBTS, RGTI, IONQ, IBM).

Promotional newsletter-style post arguing that the “most asymmetric Iran war trade” is exposure to tungsten (a critical mineral), framed as scarcity driven by geopolitical conflict and supply-chain chokepoints (Strait of Hormuz, Gulf strike) plus U.S.–China critical-minerals tensions. The post teases a “$0.30 small-cap critical mineral stock” but does not name any company or provide a tradable ticker/cashtag in the provided text.

Post argues defense stocks are at/near a bottom and set up for a multi-period upcycle because recent conflicts are driving higher defense budgets, with incremental funding skewing toward emerging technologies such as drones and counter-drone. It uses a historical analogy (Billy Mitchell/battleship-to-airpower shift) to suggest technology transitions can rapidly re-rate the winners and obsolete legacy platforms.