Are Chip Stocks Still Cheap? PEG Ratios and AI Growth Risks
The useful question is how much durable earnings growth today's share price already assumes.
Semiconductor PEG ratios can make expensive-looking stocks appear surprisingly inexpensive. A high price-to-earnings multiple looks different when earnings are expected to grow rapidly. That helps explain the interest in a chart attributed to Futurum Equities, described in the supplied research brief as showing 14 of 21 companies below a PEG of 1.
But there is a crucial distinction between a promising valuation screen and proof of undervaluation. PEG inherits every weakness in the earnings and growth estimates used to calculate it. That matters especially for memory manufacturers, where a rebound from depressed profits can produce extraordinarily low ratios near a cyclical earnings peak.
What is PEG, and how should investors calculate it?
PEG stands for price/earnings-to-growth. It divides a stock's P/E ratio by its annual earnings-per-share growth rate, expressed as a percentage number. A stock trading at 30 times earnings with expected EPS growth of 60% has a PEG of 0.5: divide 30 by 60, not by 0.60. [2]
PEG = P/E ÷ annual EPS growth in percentage points
Example: 30 ÷ 60 = 0.5
The familiar thresholds are screening conventions. Below 1 can suggest a relatively low earnings multiple for the growth assumed. Around 1 indicates numerical parity under that convention. Above 2 indicates a larger multiple relative to growth. None is a universal boundary between a good investment and a bad one.
PEG does not explicitly incorporate debt, capital expenditure, dividend payments, competitive advantage or the uncertainty of the forecast. A capital-intensive company and a licensing business can share the same PEG while producing very different cash returns for shareholders.
Similarly, a share price doubling does not establish that valuation doubled. If sustainable earnings also doubled, the P/E could be unchanged. Conversely, a falling share price can coincide with a rising multiple if expected earnings fall faster. Price performance and valuation answer different questions.
The supplied ranking: low PEG across several AI themes
The figures below come from the user's research brief, which attributes its underlying chart to Futurum Equities. They are retained for discussion, not independently recalculated. Business labels indicate broad exposure and do not imply identical revenue mixes or equal sensitivity to AI spending.
Swipe horizontally on mobile. The two missing ratios are not estimated or added to the table.
| Company | Ticker / name | Supplied PEG | Primary research question |
|---|---|---|---|
| Micron | MU | 0.1× | How durable are memory prices and margins? |
| Sandisk | SNDK | 0.2× | Can NAND earnings survive supply expansion? |
| SK hynix | SK hynix | 0.2× | What is sustainable HBM and DRAM profitability? |
| Applied Optoelectronics | AAOI | 0.2× | Will optical demand convert into profitable shipments? |
| Broadcom | AVGO | 0.3× | How do custom silicon, networking and software affect earnings quality? |
| Marvell | MRVL | 0.4× | How concentrated are customer programs? |
| NVIDIA | NVDA | 0.5× | How long can earnings growth remain exceptional? |
| Credo | CRDO | 0.5× | Will product diversification reduce customer dependence? |
| Coherent | COHR | 0.6× | How much growth reaches cash flow? |
| Taiwan Semiconductor | TSM | 0.7× | What returns follow advanced-node investment? |
| Advanced Micro Devices | AMD | 0.7× | Can product execution deliver the assumed margins? |
| Applied Materials | AMAT | 0.9× | How resilient is equipment spending through the cycle? |
| ASML | ASML | 1.0× | How do deliveries, bookings and restrictions affect growth? |
| KLA | KLAC | 1.3× | How durable is process-control demand? |
| Lam Research | LRCX | 1.3× | How much growth depends on memory investment? |
| Astera Labs | ALAB | 1.6× | Can connectivity expansion meet high expectations? |
| Arista Networks | ANET | 1.7× | Will customer spending and networking margins hold? |
| Intel | INTC | 1.8× | Is projected EPS recovery economically sustainable? |
| Arm | ARM | 2.5× | How much future royalty growth is already priced in? |
Within these 19 disclosed values, 12 are below 1, one equals 1 and six exceed 1. The brief's broader 14-of-21 claim equals 66.7%, but it cannot be fully reconstructed from the supplied values alone. Rounding also matters: a displayed 1.0 might conceal an underlying figure just above or below that threshold.
Even if the complete count is accurate, it describes a selected group rather than the whole semiconductor industry. Arista is a networking systems company, while optical suppliers, equipment makers and chip designers occupy different parts of the value chain. The selection is best treated as an AI infrastructure watchlist.
The first question is which earnings and which growth rate
A PEG ratio is only comparable when its inputs are comparable. Some providers use trailing earnings; others use forward estimates. Growth may mean the next fiscal year's increase or a three-to-five-year annualized forecast. Fidelity's research glossary, for example, defines PEG using forward P/E and projected three-to-five-year annual EPS growth. [3]
A 100% rebound in next year's earnings is not equivalent to 100% compounded annually for five years. The latter would imply earnings increasing thirty-twofold over the period. Using an exceptional one-year rebound as if it were a sustainable multi-year rate dramatically changes the interpretation.
Accounting conventions matter just as much. GAAP EPS and adjusted EPS can differ because of stock-based compensation, acquisition charges, restructuring and investment gains. Dividing a GAAP P/E by growth calculated from adjusted earnings creates a ratio with inconsistent inputs.
Every useful comparison should identify the share-price date, EPS basis, forecast horizon, growth method and data provider. Fiscal calendars also differ across companies. Without those details, a precise-looking ranking can create more confidence than the underlying estimates justify.
Micron's low PEG: strong results and a difficult denominator
Micron's September 30 fiscal fourth-quarter 2026 release confirms non-GAAP gross margin of 87.0%, compared with GAAP gross margin of 86.8%. Revenue reached $54.229 billion and adjusted diluted EPS was $33.42. Those are reported results; the 0.1 PEG in the supplied brief is a separate valuation estimate. [4]
Micron's same release guides next-quarter non-GAAP gross margin to approximately 86.25%. That remains a high level, but it illustrates why the current quarter should not be treated as a permanent operating condition. Management's demand outlook and actual future profitability are different types of evidence. [4]
Memory profits respond to shipment volumes, product mix, pricing and manufacturing costs. Tight supply can lift prices faster than costs, creating rapid EPS growth. A tiny PEG may therefore reflect the steepest portion of the earnings cycle rather than a uniquely inexpensive long-term business.
Consider the brief's hypothetical path: EPS rises from $10 to $30, then falls to $15. The initial increase is 200%, followed by a 50% decline. Across the two years, annualized EPS growth is approximately 22.5%, calculated from the starting and ending earnings. The peak year's percentage tells a very different story.
For Micron, Sandisk and SK hynix, examine normal-cycle margins, planned capacity additions, customer inventories and contract protections. Long-term agreements may improve visibility, but their pricing formulas, minimum purchase requirements and cancellation clauses determine how much risk they actually remove.
NVIDIA: can earnings keep catching up with the price?
NVIDIA's August 26 fiscal second-quarter 2027 release reported revenue of $96.2 billion, up 106% year over year. Its filings describe strong AI and accelerated-computing demand while identifying competition, manufacturing dependencies and product-transition risks. These disclosures support analysis of growth, but they do not independently establish the supplied 0.5 PEG. [5]
The central valuation question is how long earnings can compound at the rate embedded in the model. A product roadmap can expand the opportunity, yet future launches still require customer adoption, available infrastructure, effective deployment and attractive economics for buyers.
Strong revenue growth also need not translate into proportionate EPS growth. Gross margins, research spending, taxes and the share count affect the result. A model assuming both rapid sales expansion and permanently exceptional margins deserves separate tests for each assumption.
In a hypothetical example, a stock at 30 times earnings would fall to 20 times those future earnings if EPS rose 50% while the price stayed unchanged. That is earnings catching up with valuation. It is a possible outcome, not a promise that investors will enjoy both the growth and an unchanged multiple.
AI spending reaches more than GPU suppliers
The brief highlights a useful industrial distinction. Computing requires memory, data movement, storage and manufacturing capability. A successful AI deployment can therefore create opportunities across several suppliers, even when one chip designer dominates the public conversation.
Photonics and connectivity
Optical components and high-speed links help move information between systems. For suppliers such as AAOI, Coherent, Credo and Marvell, the research task is to trace that need into qualified products, firm orders, manufacturing yields and cash collection. A growing addressable market is not the same as profitable market share.
Small profit bases can also exaggerate percentage growth. EPS rising from $0.10 to $1.00 produces 900% growth, but says little about what comes next. Customer concentration and the timing of individual deployment programs can make forecasts especially sensitive to a few decisions.
Foundries and semiconductor equipment
TSMC, ASML, Applied Materials, KLA and Lam Research occupy different positions in chip manufacturing. Their earnings depend on technology investment, utilization, product mix and delivery schedules. A higher PEG than a memory producer may reflect steadier growth rather than an inferior business.
Equipment orders can lead revenue, while customer capital budgets can change before deliveries occur. Backlog quality, export restrictions and service revenue deserve attention alongside EPS. Compare companies with similar economic roles before interpreting a cross-industry ranking as an investment priority list.
What do high PEG ratios say about Arm and Intel?
The supplied table places Arm at 2.5 and Intel at 1.8. These numbers imply different P/E-to-growth relationships under the unknown source methodology. They do not establish that Arm must fall, that Intel is safer, or that either ratio accurately describes today's valuation.
A licensing business with strong incremental margins can merit a different valuation from a manufacturer funding a major turnaround. Investors should examine earnings durability, cash conversion and required reinvestment rather than demanding the same PEG from every business model.
For turnaround companies, PEG can become unreliable near zero earnings. Moving from a loss to a profit does not produce an ordinary economically meaningful percentage growth rate. Restructuring benefits and unusually weak comparison periods can also distort the apparent recovery.
The higher-PEG names therefore belong on a different research track. Ask what growth and margins the price requires, then compare those requirements with reasonable scenarios. A premium valuation raises the consequences of disappointment; it does not prove that disappointment will occur.
How fragile is a low PEG?
The following example holds P/E at 30 and changes only the assumed annual EPS growth rate. It shows how a revision to the growth denominator can change the screen without any share-price move.
| Annual EPS growth assumption | Calculation | PEG | Interpretation |
|---|---|---|---|
| 60% | 30 ÷ 60 | 0.50 | Looks low if that growth is credible |
| 40% | 30 ÷ 40 | 0.75 | Still below the conventional threshold |
| 30% | 30 ÷ 30 | 1.00 | Numerical parity, not proven fair value |
| 20% | 30 ÷ 20 | 1.50 | More demanding relative to growth |
| 10% | 30 ÷ 10 | 3.00 | The original apparent discount disappears |
In reality, a forecast cut can affect both inputs. If forward EPS falls, forward P/E rises at the same share price. If the projected growth rate also falls, PEG increases again. This double effect explains why the metric can deteriorate rapidly during an earnings revision cycle.
Low PEG does not prove that earnings have already grown faster than the share price. That historical claim requires comparing actual earnings and price changes over matching dates. A forward PEG principally describes how today's valuation relates to an assumed future growth rate.
A practical way to use PEG in stock research
Start with a consistent screen, then replace the headline ratio with business-specific questions. The objective is to identify where further work may be rewarding, not to buy every company below an arbitrary threshold.
| Research step | What to check | Why it matters |
|---|---|---|
| Standardize inputs | Price date, EPS basis and growth horizon | Prevents misleading comparisons |
| Normalize earnings | Cycle position, one-offs and weak bases | Tests whether low PEG is temporary |
| Follow cash | Operating cash, capex and working capital | Measures reinvestment demands |
| Inspect concentration | Customers, geography and suppliers | Reveals dependence behind forecasts |
| Stress the model | Slower growth and lower margins | Tests the downside before investing |
Translate growth into an expected return
A simple hypothetical shows why PEG should be followed by a valuation model. Suppose a stock costs $100 and earns $5 per share. If EPS compounds at 20% for three years, it reaches $8.64. At an eventual P/E of 20, the share price would be $172.80, before dividends. At a P/E of 12, it would be $103.68.
The same earnings growth produces radically different investment outcomes because the exit multiple changes. Neither multiple is a prediction. The exercise makes the dependence visible and allows investors to ask whether the expected return remains acceptable if enthusiasm fades.
Then examine cash requirements. Earnings growth funded by large factory investments, acquisitions or continuing share issuance may deliver less cash per share than the income statement initially suggests. Conversely, cash generated from existing assets can support investment without repeated external funding. Compare free cash flow over several years rather than assuming a single strong quarter captures the business's normal economics.
The goal is a range of plausible returns, with explicit assumptions that can be updated after each earnings report.
Also consider portfolio overlap. Owning a GPU supplier, a memory producer and an optical company can still represent one concentrated bet on data-center spending. Different ticker symbols do not guarantee independent earnings risks.
Frequently asked questions
Does PEG below 1 mean a stock is undervalued?
It may flag an interesting relationship between valuation and growth, but does not establish fair value. Forecast quality, the earnings base, balance-sheet risk and required reinvestment can overturn the initial impression.
Why can memory stocks have extremely low PEG ratios?
Rapid profit recovery and rising memory prices can create unusually high EPS growth. That large denominator reduces PEG, even when the growth rate is unlikely to persist through a full industry cycle.
Are all 21 chart values independently verified?
No. The supplied brief lists 19 explicit ratios. The original forecast dataset and calculation methodology were unavailable, so this article does not certify the complete 14-of-21 count or present the ratios as current quotes.
Is a higher PEG always worse?
No. A company with durable earnings and limited reinvestment needs can deserve a premium over a cyclical business. PEG alone cannot measure the stability or economic quality of growth.
Can PEG be used when a company is losing money?
Usually not in a meaningful conventional way. Negative earnings or growth from a near-zero base can make the ratio unstable. Analyze the path to sustainable profitability and cash generation instead.
Chip stocks can rise sharply without becoming uniformly expensive. But low PEG is an invitation to examine the growth assumptions. The strongest investment case combines a reasonable price with earnings that can survive competition, reinvestment requirements and the next turn in the cycle.
Sources and methodology
Research date: October 7, 2026, Bangkok time. Company results are sourced from official releases. The valuation table reproduces the supplied brief and is not a refreshed consensus dataset. All numerical sensitivity examples are hypothetical. No unseen chart values have been invented.
- Outlier Capital: secondary repost of a chart attributed to Futurum Equities. Establishes circulation of a related chart, not verified forecast inputs.
- Fidelity: company valuation ratios.
- Fidelity research glossary: PEG methodology.
- Micron: fiscal fourth-quarter and full-year 2026 results.
- NVIDIA: fiscal second-quarter 2027 results; related SEC Form 10-Q.
Disclaimer: Educational information only, not personalized investment advice. Estimates can change, investments can lose value, and a favorable valuation ratio does not guarantee future returns.