MPWR · Monolithic Power Systems, Inc. · Stock research

Current Thesis

The July binary is settled and the market has spent three weeks unwinding the reaction to it. Q2 2026 revenue of $980.6M against a $890–910M company guide, Enterprise Data of $380.6M (+164.3% YoY, 38.8% of revenue) and a Q3 guide of $1,140–1,160M versus a Street number near $982M were all delivered on 2026-07-30. The high-water mark of the response was the 2026-07-31 pre-market quote of $1,465.50. The 2026-08-21 close of $1,316.28 sits below that, 22.0% under the $1,687.32 52-week high, with a three-month price change of -17.1% and RSI(14) at 46.4.

What is new since the last note is a governance shock, not a demand one. On 2026-08-18 the shares fell $97.13, from $1,415.99 to $1,318.86 — a 6.86% session with intraday selling reaching -7.61%. Weiss Ratings' same-day alert attributed the move to a report of a shareholder investigation into possible breaches of fiduciary duty by officers and directors, and noted the report named no law firm, no filed complaint and no alleged financial misstatement. That decline carried price back to the 2026-06-26 pivot low of $1,313.32; the 2026-08-21 weekly close of $1,316.28 is roughly two-tenths of a percent above it.

The narrative is saturated. The story still works fundamentally — the 2026-07-30 numbers accelerated — but the bid behind it has thinned to the point where a 16%-above-consensus guide bought no new high and a routine litigation-fishing headline moved the stock nearly 7% in a session. Dating the label: the 07-31 gap fully retraced by 08-18; the 2026-04-24 launch of a 2X leveraged single-stock MPWR ETF (Tradr); and Benzinga's "if you invested $1,000 20 years ago" retrospective on 2026-08-21, the third such backward-looking return piece since 2026-04-01. Coverage of that shape clusters after attention peaks, not before it.

Bullish and bearish views on Monolithic Power Systems, Inc.

The model's bull view on Monolithic Power Systems, Inc. (MPWR), in brief: Q2 2026 revenue $980.6M (release 2026-07-30), above the $890–910M guide issued 2026-04-30 and the ~$902.5M consensus; +47.6% YoY and +21.9% QoQ, a record quarter. The bear view: The governance file reopened. The 2026-08-18 session lost 6.86% on a shareholder-investigation report with no named plaintiff firm and no complaint attached. That the market paid that much for a low-information headline says the holder base is not committed. The investigation… Both cases follow in full.

Bull Case

• Q2 2026 revenue $980.6M (release 2026-07-30), above the $890–910M guide issued 2026-04-30 and the ~$902.5M consensus; +47.6% YoY and +21.9% QoQ, a record quarter.
• Enterprise Data $380.6M, +164.3% YoY and +44.8% QoQ, 38.8% of revenue versus 32.7% in Q1 2026. This is the direct numerical answer to the November-2024 claim — still being litigated in the open law-firm investigations — that MPWR lost vertical-power sockets on NVIDIA accelerator boards.
• Q3 2026 guide $1,140–1,160M against a Street estimate near $982M. A raise of that magnitude is a content-and-share statement about the current accelerator generation.
• Mix shift has not cost margin: Q2 GAAP gross margin 55.2%, non-GAAP 55.6%, with Q3 non-GAAP GM guided 55.4–56.0% and non-GAAP opex $201.2–205.2M.
• Non-GAAP EPS $6.50 vs $5.87 consensus; GAAP operating income $303.9M (2026-07-30).
• Buyback stepped up into the drawdown: the board added $500M to the repurchase authorization on 2026-07-30, taking it to $1B, against $1,413.8M in cash and short-term investments.
• A Barchart summary of an S&P Global poll of 15 analysts put the consensus at Strong Buy with an average target of $1,795 — all of these above the 2026-08-21 close.

Bear Case

• The governance file reopened. The 2026-08-18 session lost 6.86% on a shareholder-investigation report with no named plaintiff firm and no complaint attached. That the market paid that much for a low-information headline says the holder base is not committed.
• The investigation drumbeat is continuous and dated: the Schall Law Firm announced a fiduciary-duty investigation on 2026-06-03, Johnson Fistel on 2026-07-28, with further reports carried on 2026-08-18. Kahn Swick & Foti's stated predicate is the report that NVIDIA cancelled half its orders over product performance. Law360 has separately carried coverage of a pending investor fraud suit in which executives argue their own stock sales rebut the claims; the docket date is not established here.
• The material weakness behind the 2026-02-26 non-reliance 8-K remains unremediated, still named as a risk in the 2026-07-30 release. Every governance headline lands on that unresolved base.
• The best print in the company's history did not produce a new high. Price gapped to $1,465.50 on 2026-07-31 and closed 2026-08-21 at $1,316.28.
• Insider flow is one-directional: Form 4 records showed roughly 127 sales and zero open-market purchases over the six months to 2026-07-18, including CEO Hsing at ~194,923 shares and GC Tseng at ~79,113 shares.
• Concentration: 38.8% of revenue on one AI-accelerator power line tied largely to one architecture generation. A single design-win loss at the next platform transition takes the growth rate with it.
• The Q3 bar is now self-set at $1,140–1,160M. Guidance that far above consensus removes the cushion; the October print has to clear a number the company chose.

Setup & Price Structure

The 2026-08-21 close of $1,316.28 is the reference level. Structure below it is thin: the 2026-06-26 pivot low of $1,313.32 is the last shelf before the ~$1,200 area that framed the spring range, and the weekly close is sitting on it rather than above it. RSI(14) at 46.4 is mid-range — there is no oversold washout here, so the decline has been orderly rather than capitulatory.

Overhead, three levels matter in sequence. The 2026-08-17 close of $1,415.99 is the origin of the 08-18 gap-down and the first thing a recovery has to fill. The $1,465–1,470 zone contains both the 2026-07-31 post-print pre-market quote and the mid-July trading area; reclaiming it on weekly closes is what would convert the July numbers into a price trend. The $1,687.32 52-week high is 22.0% away and not in play on this leg.

Positioning observables, stated as observables: the 2X leveraged single-stock ETF that launched 2026-04-24 adds mechanical daily rebalancing flow around volatility events, which cuts both directions; retail-facing retrospective-return coverage has appeared three times since 2026-04-01, most recently 2026-08-21; insider sales run 127-to-zero over the trailing six months to 2026-07-18. Price extension is not part of the crowding picture — the stock is below where it traded three months ago. The crowding is in attention and product wrappers, while the price has already de-rated.

No company-specific event falls inside the next 30 days. The nearest binary is external.

Catalyst Calendar (next 30 days)

• 2026-08-26 — NVIDIA Q2 FY2027 results and call (quarter ended 2026-07-26). The dominant read-across for MPWR's Enterprise Data ramp. Data-center revenue, next-generation platform commentary and any power-delivery architecture detail set the tone for the accelerator-power supply chain.
• No scheduled MPWR filing, conference or guidance event is on the calendar between 2026-08-23 and 2026-09-22. Any move in the window is read-across or headline-driven.
• ~2026-10-29 (est.) — MPWR Q3 2026 results. First test of the $1,140–1,160M guide and of whether Enterprise Data extends sequentially after +44.8% QoQ.
• ~2026-10-30 (est.) — Q3 2026 10-Q. The next scheduled document that can update the internal-controls conclusion carried since 2026-02-26.

What Would Change Our Mind

The structure that has held every drawdown since June is the 2026-06-26 pivot at $1,313.32, and the 2026-08-21 weekly close of $1,316.28 leaves almost no room in it. A weekly close below $1,313 forfeits that pivot and the post-print base with it, opening the ~$1,200 shelf; at that point the read becomes that the market is repricing governance and concentration risk faster than the income statement can compound, and the July beat stops being the relevant input.

Three other things would flip the read without a price break. First, NVIDIA's 2026-08-26 result passing with strong data-center numbers while MPWR fails to recover the $1,415.99 gap origin — that would isolate the weakness as company-specific rather than sector beta. Second, any escalation from investigation notice to a filed derivative or securities complaint naming specific transactions, or a Q3 10-Q that repeats the material-weakness conclusion. Third, Enterprise Data declining sequentially at the ~2026-10-29 print, which would end the content-ramp story on its own terms.

The opposite case is equally observable: weekly closes reclaimed above $1,470 with Enterprise Data extending in October would mean the August de-rating was a governance-headline air pocket inside an intact ramp.

Correlation Notes

• NVIDIA is the primary correlate. Enterprise Data at 38.8% of revenue is levered to accelerator build rates, and the 2026-08-26 print is the near-term driver of MPWR's tape.
• SOXX / SMH. The April 2026 chip-ETF melt-up (SOXX +27% in April, record combined inflows of $5.45B into SOXX and SMH, reported 2026-04-21) and the subsequent reversal have dominated MPWR's price path more than its own results. Sector beta is currently the larger term.
• Infineon and Renesas are the named alternative vertical-power suppliers in the 2024 socket-loss claim. Any disclosure at a platform transition that shifts power-delivery content toward them is a direct MPWR negative.
• The 2X single-stock ETF (Tradr, launched 2026-04-24) introduces daily rebalancing flow that amplifies both directions around high-volatility sessions such as 2026-08-18.
• Analog and power peers (ADI, TXN, Vicor) share the industrial/automotive cycle exposure but not the AI-accelerator concentration; divergence between MPWR and that group isolates how much of any move is data-center-specific.