WickedDesk Weekly #003: Robotics After The Selloff

The short US week effectively ends on Thursday. NYSE and Nasdaq are closed on Friday for the Independence Day holiday, so Thursday close is the key input for next week's preparation. On the week, the large indices were firm, and the second quarter closed as the strongest since 2020.

The break came Wednesday and Thursday, and the trigger was not the macro calendar. Meta said it would rent out excess compute capacity (Meta Compute), which questioned the scarcity assumption behind much of AI hardware valuation. Selling ran down that chain, semis and memory first. Micron fell sharply despite strong numbers, and the Asian memory names came under pressure at the same time.

The jobs report was weaker than expected: 57,000 new jobs, unemployment at 4.2%, wages +3.5% year over year. For rate expectations that was relief rather than the reason for the selloff.

In the rotation view Technology lost 2.7%, Industrials lost 1.0%. Healthcare, Staples, Real Estate, Materials and Utilities held better, while Communication Services and Financials led the week. That does not remove robotics from the list, but it narrows the work. This issue checks four roles: Machine Vision with CGNX, industrial automation with ROK, Edge Vision AI with AMBA and force sensing with VPG. From Monday on, the question is whether those names find buyers after the selloff or whether rotation keeps moving toward Healthcare and more defensive groups.

This week had fewer trading days, but it gave clear inputs for the next plan. The holiday shortens confirmation. The real driver came from AI infrastructure: Meta saying it would rent out excess compute capacity broke the scarcity assumption and pulled semis, memory and robotics lower on Wednesday and Thursday.

The jobs report changed rate expectations, but it was the accompaniment, not the trigger. This section is not a calendar recap. The useful question is what changes the work: Technology and Industrials need to stabilize, Healthcare remains visible as a comparison group, and robotics needs a narrower list than it had before the selloff.

Thursday was not an even decline. The weakest blocks were Technology and Industrials. Many recent watchlist names were hit there: OUST, ACMR, MXL, VPG, SNDK, COHU, PENG, AMAT, LRCX, MRVL, AMBA, NBIS, CRDO, DELL.

The better areas were Healthcare, Staples, Real Estate, Materials and Utilities. Names such as MRNA, PGEN, CELH, SAM, BAX, AUGO, SSRM, PLMR appeared in those blocks. That is not a reason to abandon growth. It is a reason to sort the next list by stability first.

Monday has two paths. Either Technology reclaims part of the drop and robotics stays active. Or Healthcare and more defensive groups keep leading, and the study list shifts there.

The Matrix shows which stocks still sit close to 52-week highs after Thursday. NAVN, PTRN are directly at the high. MBX is roughly 1% below. Then come LQDA, OSCR, TXG, ABVX, SEZL, ARWR, DAVE.

Healthcare stands out: ABVX, MRNA, PGEN, FTRE, RCUS, ELVN remain near the front. Robotics, semis and hardware look different. VPG, LRCX, COHU, PENG, AMAT, LPTH, CEVA sit much deeper. These names are not removed, but they need a clean stabilization attempt first.

The Matrix separates two jobs: the strongest names near highs stay on the first check list. Robotics remains a dedicated study chapter, but not every robotics supplier gets the same rank.

Group Strength sets the frame. Robotics does not lead the list, Technology was sold on Thursday in the AI repricing, and Industrials also weakened. So the theme is read more narrowly: Machine Vision, Edge Vision AI, industrial automation and force sensing.

The better starting points have clear roles. CGNX is Machine Vision, AMBA is Edge Vision AI, ROK is industrial automation, VPG is force sensing. Healthcare remains the most important comparison group because more names held near highs on the weak day.

If CGNX and ROK stay stable while AMBA and VPG pull back in a controlled way, robotics stays on the list. If not, the focus moves to the Matrix leaders and stronger groups.

Robotics is not one ticker and not one ETF. The value chain runs from compute to sensing, motion control, manufacturing and software integration. Different stocks answer different questions: NVDA and QCOM for compute, AMBA for edge vision, CGNX for machine vision, ROK for industrial automation, RRX for motion control, VPG for force sensing and JBL for manufacturing.

Thursday was the first stress test. OUST, VPG, PENG, AMBA were sold hard. That does not make them irrelevant, but it changes the assignment. The next check is stabilization, a smaller daily range and relative strength against their own basket.

The Builder is available to every subscriber. It follows an approach inspired by Leopold Aschenbrenner: ask the right question first, then map bottlenecks, supply chains and counterarguments, and only then move to individual stocks.

For this issue, the question was: which listed companies benefit from humanoid robotics, machine vision and industrial automation before the end products sell at scale. The Builder breaks the chain into compute, camera, sensing, gearing, motion control, manufacturing, software and integration. Then it creates a thesis and an antithesis. Thesis: robotics may show up through suppliers before end-product makers. Antithesis: many revenues are still prototype revenues and may not scale linearly.

That means AMBA, CGNX, ROK, RRX, VPG, JBL, NVDA, QCOM do not belong in one bucket. Each name gets a role, a check and a reason why it can leave the list again.

Humanoid robotics is still early as an end market. That makes the supplier chain more important than the broad headline. A system needs compute, cameras, force sensors, precision gears, motors, motion control, batteries, manufacturing and software integration.

Three roles matter for this issue. RRX supplies motion-control components, motors and gearing that are relevant for joints and movement. VPG supplies force sensing, strain gauges, load cells and precision resistors. These parts help a system measure force, grip and react safely. Harmonic Drive remains an important precision-gear context, but the Atlas list starts with US-listed names.

The limitation belongs next to the opportunity: many revenues still come from prototypes, not production scale. So this issue does not ask who has the biggest headline. It asks which chart finds buyers first after a weak Thursday.

13F data does not show what was traded today. The Duquesne filing shows holdings as of 2026-03-31 and was filed in mid-May. It is not a trading plan. It is useful as an idea filter because it shows where a large thematic investor had capital clustered.

After Thursday, the comparison matters most for Healthcare. NTRA, INSM, NAMS, OPCH sit in an area that led on the weak day. Tech and semi positions such as TSM, STM, AVGO belong on the reclaim list. Country and materials ideas such as EWZ, YPF, ARGT, AA show that capital is also working outside the classic AI names.

This issue turns that into a work list, not a recommendation: check Healthcare against the Matrix, check semis for reclaim, and only include country or materials ideas if the group confirms.

The Builder does not start with a stock. It starts with a precise question. In robotics, the first output is a set of roles: platform, compute, sensing, vision, motion control, manufacturing and software. Then it checks which listed companies are visible in those roles.

The important part is the limitation. If a company only has a small prototype shipment, the headline is not enough. If the stock is far below its high, the theme is not enough either. A candidate needs role, data point, chart and group to line up.

BOTZ, ROBO, ARKQ show the broader robotics and autonomy basket. The check answers a simple question: does the theme have several working names, or is it only one ticker.

The current read is mixed. All three are positive over three months, but all sit well below 52-week highs. That is not enough for a broad robotics list. It is enough for a narrower question: which individual names hold better than the basket. In this issue, that starts with CGNX, ROK. AMBA, VPG, PENG, OUST need visible stabilization after Thursday.

After Thursday, the headline matters less than the distance to the high. NAVN, PTRN are directly at the high. MBX, LQDA, OSCR, TXG sit just below. These are the first names for the cross-check because they did not build much distance despite a weak growth day.

Robotics looks different. CGNX, ROK remain workable. AMBA, VPG, PENG, OUST need stabilization first. That difference matters: the theme is interesting, but the candidates do not get equal treatment.

The study list stays small. CGNX is the cleanest robotics anchor because machine vision is industrially concrete and the chart remains closer to the high than many speculative names. ROK adds the more mature automation block. AMBA remains the momentum name from Edge Vision AI, but only with a tighter range after the pullback.

VPG is deliberately included. Force sensing is not a side product for humanoid robots. It is required for gripping, load measurement, torque control and collision avoidance. The company supplies strain gauges, load cells, force sensors and precision resistors. The drawback is just as clear: the market sold VPG hard on Thursday. That makes VPG a pullback study with a strict check, not a breakout.

The Atlas check has four questions. First: does CGNX keep proximity to the high while Technology swings. Second: does ROK stay steadier than the industrial block. Third: does AMBA build a controlled range after the rally instead of falling back into the old area. Fourth: does VPG find buyers after the selloff without immediately losing Thursday’s range.

If two or three of those questions are answered positively, robotics stays active next week. If not, the theme is parked and the work moves to the Matrix leaders.

AMBA represents Edge Vision AI in this issue. The appeal is not a slogan. The role matters: camera data must be processed locally when robots need to react quickly. That is where Ambarella’s vision processors fit.

The chart is no longer early after the rally. Thursday showed how quickly momentum names can be sold under pressure. For next week, one green day is not enough. The cleaner setup would be a smaller range, a higher low and demand above the area where the rally began.

VPG is not a classic robotics hype stock. Its focus is measurement: force, weight, strain, torque and precision. For humanoid systems, that matters because a robot needs to measure force while gripping and moving, not only see its surroundings.

The fundamental limitation is clear: humanoid robotics revenue is still small and can fluctuate. That is why VPG is a study candidate, not a core name. The stock has RS 99, but it was sold roughly 13% on Thursday and sits well below the high in the Matrix. The plan is simple: stabilization first, then a new base. Without that, VPG stays on watch.

CGNX is the steadier but more important robotics name in this issue. Machine Vision is not a future promise. It is an existing industrial building block. Cognex supplies identification, recognition and inspection in manufacturing, which connects to robotics without depending only on humanoid end products.

ROK is the more mature automation name. The FactoryTalk Orchestration launch at Automate 2026 fits the theme, but the chart decides whether the stock remains only context or stays on the study list. Both stocks are less spectacular than AMBA or VPG. That is exactly why they are the better first checks after the selloff.

OUST, SERV, RR do not automatically belong in the front list just because they look more directly tied to robotics. After Thursday, distance to the high and follow-through quality matter more.

OUST was sold hard. SERV and RR remain very speculative and sit far below their highs. These names can stay on a separate watchlist, but they should not replace the work on stronger suppliers.

PATH, ISRG, HON, SERV, RR stay outside the front study list for different reasons. PATH fits the automation theme but does not have current leadership in the chart. ISRG is high quality but sits far below the high. HON is too broad and does not lead in the chart. SERV, RR are too speculative for the main list.

The point of exclusions is not negativity. It keeps the list from filling with names that fit the theme but do not fit the current market well enough.

The most important driver of the short week did not come from the macro calendar but from AI infrastructure. Meta renting out excess compute capacity questioned the scarcity assumption behind AI hardware valuation and hit semis and memory first. Micron fell despite strong numbers, and the Asian memory names came under pressure at the same time. For balance, the BIS warned about stretched AI valuations.

The jobs report sat underneath that: 57,000 new jobs were clearly below expectations, the unemployment rate held at 4.2%, below consensus, and wages rose 3.5% year over year. For rate expectations that was relief. Oil eased again on the week alongside progress in the US-Iran talks, which takes pressure off the inflation side.

For stocks, the message is split. A friendly macro input helps, but Thursday shows that growth groups can still be sold in a repricing. The key question is which groups attract capital on Monday and which ones only bounce briefly.

Monday decides whether Thursday was a hard pullback in leading areas or the start of rotation away from Technology. The first hour belongs to groups, not favorite individual charts: Technology, Industrials, Healthcare, Staples and Financials.

Positive scenario: Technology reclaims part of the decline, semis stop selling off, CGNX and ROK hold better than their groups, and AMBA plus VPG build tighter follow-up days. Then robotics stays high on the study list.

Negative scenario: Technology and Industrials keep losing ground, OUST, AMBA, VPG, PENG and semis make new lows, while Healthcare and Staples continue to lead. Then the weekly list changes: less robotics, more Matrix leaders and Healthcare names.

Three lists are prepared over the weekend. List one: Matrix names near highs, especially NAVN, PTRN, MBX, LQDA, OSCR, TXG and Healthcare. List two: the robotics chain with CGNX, ROK, AMBA, VPG, RRX, PENG, OUST, JBL. List three: 13F ideas that match current rotation.

Each name needs only three answers: Which group supports it. How far is it from the high. What invalidates the setup. Everything else stays in the archive.

A large theme becomes useful when it is broken into roles. Robotics is too broad. Sensing, edge vision, machine vision, motion control, manufacturing and software are easier to check.

The practical flow: build the chain first, assign the stocks, then check Matrix and Atlas. That does not create a long idea list. It creates a short list with clear reasons. For this issue, the working names are CGNX, ROK, AMBA, VPG.