Vishay Precision Group

The Verdict

We rate Vishay Precision unattractive. It makes ultra-stable electronic parts and force sensors, and chip-equipment makers are ordering at a record pace. The business is mixed: the orders have not reached profit, because head-office and new costs took nearly everything its businesses earned. The price assumes a profit margin by 2028 above anything it has earned since listing in 2010, so it is stretched, priced for perfection. As a bet on humanoid-robot sensors it is fragile: if that line hits management's goal, the shares would still be worth less than today; if it fails, a little over half is left.

The long caseIf humanoid robots leave the lab for factory floors in large numbers, each needs precise force sensing, and a supplier already nominated by one robot maker could ride that growth.

The company

VPG takes a metal foil a few thousandths of a millimetre thick and etches it into two kinds of part. One is a resistor that sets an exact voltage in a measuring circuit and barely drifts as it warms. The other is a strain gauge: glued to metal, it measures how hard the metal is being pushed.

Chip-test equipment, aircraft and satellites use the resistors; robot joints, scales and truck-weighing systems use the gauges. VPG also builds finished load cells, steel-mill gauges and crash-test recorders.

How it makes money

It sells parts and systems, one order at a time, through three businesses of similar size: Sensors ($115.6M of 2025 revenue), Weighing Solutions ($111.1M) and Measurement Systems ($80.4M).

Each business earns a profit, but a large head office sits on top. In 2025 the businesses earned $45.2M and head-office costs took $35.4M, leaving $8.3M of operating profit once a one-off gain from selling a factory is removed.

Who runs it

Ziv Shoshani has run VPG since it split from Vishay in 2010, and the same finance chief has been beside him since then. Profit margins peaked in 2022 and have since shrunk. Insiders own about 7% of the shares but, through a family trust's super-voting stock, about 29% of the vote. Bonuses ride on adjusted profit and margin, the long-term shares on earnings and cash, and neither the bonus nor those shares have paid out lately.

Why it wins, and who could beat it

Why customers come back

A designer who needs a circuit to stay accurate as temperatures swing writes that exact foil part into the design. Changing it later means proving the circuit again, and for military and space parts, requalifying it.

The edge is real at the top: VPG's qualified military resistor drifts at most 2 parts per million per degree, while the best space-approved rival part is rated at 5. Below that, cheaper parts will do: Sensors kept about 31 cents of each sales dollar as gross profit in Q2 2026.

The threat

In robots, the attacker is the finished sensor. Novanta sells ready-made force sensors and names humanoids as a market, so a robot maker can buy the whole unit and never order VPG's gauges.

In resistors, Vishay Intertechnology's thin-film parts keep getting more precise, and in load cells and steel-mill gauges VPG competes with HBK, Zemic, ABB and others, where price decides.

The rivals: NOVT, VSH, and 2 more that are private, in-house or listed abroad.

How big it could get

How big this could get

The humanoid market is still tiny. An industry body counts about 7,000 humanoid robots sold worldwide in 2025, many bought for research. At the $400 to $500 of sensing parts management says each robot carries, that is about $3M, less than the roughly $4M VPG says it sold to robot makers that year.

So today's humanoid sales are prototypes. At volume, management expects $150 to $250 a robot, so thousands a week are needed to matter to a $330M company.

Who has signed

The first humanoid customer sent a formal supplier nomination, which names VPG but commits to no volume or price. Across the company, $135.8M of orders were waiting to ship in July 2026, $78.3M of them in Sensors, though customers can often cancel without penalty.

The results

The numbers

The last report (5 Aug)

Quarter ended July 2026, filed figures.

Q2 2026 sales rose 11.7% to $83.9M, below the $85M to $90M the company had promised, and the company made an operating loss of $0.3M.

The problem was not demand. Sensors revenue rose 25.8% and its profit rose 1.2%, because VPG hired and paid more ahead of the orders, and head-office costs rose too. Management blamed a $3.3M currency hit and $3.0M of shipments delayed by a new computer system.

Humanoid sales were $0.32M, about half of Q1, after management had said they could double.

What it did to the thesisIt weakened the case for the plan the price already pays for: orders grew but did not turn into profit, the company fell short of its own forecast, and humanoid sales fell when they were meant to rise. That is why the view stays Unattractive.

The price, and where we disagree

In short. Priced for perfection. The price pays in full for a profit margin VPG has never reached, before its new orders have added any profit.

The shares tripled in a year to $151.78 on the robot story, then fell 27.2% on the day of the Q2 2026 results. At $71.63 they are still more than double their level of a year ago.

Chip-equipment spending is at a record, and the $135.8M of waiting orders should show in Q3 2026 results on 3 November. Short sellers hold 13.8% of the shares available to trade, so the market is split.

Where we disagree with the market

Analysts' 2028 forecasts are simply the company's plan, and every analyst rates the shares a buy. They read the orders. The accounts for each business show the extra revenue being spent on new staff, materials and head office before it becomes profit.

The evidence against us is management's own sum: without currency, an unusual mix of products and the delayed shipments, Q2 2026 would have earned about $9M before interest, tax and depreciation rather than $5.5M.

What could go wrong, and how we would know

The risks

• We may be early. If currency and the delayed shipments explain the lost profit, Q3 and Q4 2026 could show the jump management expects.
• A real robot ramp. The first humanoid customer plans hundreds to thousands of robots a week; if it happens, sales could beat any plan.
• The chip cycle turns. Test-equipment sales fell by a fifth in 2024; a downturn would drain the waiting orders the case rests on.
• Factories in Israel. Its main plants are there, so war or a stronger shekel hits output and costs directly.
• A family controls the vote. A trust run partly by the chief executive holds 27.4% of the votes, through ten-vote shares.

What would change our view

1. Unattractive→AttractiveThe margin arrives, at a price short of assuming it

   It needs all of:

   • ○The operating margin, before one-off costs, reaches 8% for two quarters in a row.checked at the next results, due by 3 Nov
   • ○The whole business costs no more than 18 times its eight-year average operating profit.about ≈$41.86 today, 29% below the last close
2. Unattractive→NeutralOperating profit follows the orders

   It needs:

   • ○Sensors profit grows at least as fast as Sensors revenue for a year, and head-office costs fall below 10% of revenue.checked at the next results, due by 3 Nov
3. Reopens the researchThe humanoid ramp is contracted

   It needs:

   • ○VPG discloses a humanoid order with a volume and a price, or humanoid sales pass $2M in a quarter.watched as it happens

What is coming

• 3 November 2026 Q3 2026 results: Whether record backlog turns into profit: adjusted operating margin against 5%, Sensors profit against Sensors revenue, humanoid sales against Q2's $0.32M.