That's how much revenue NVIDIA generated from its Data Center business in a single quarter, an extraordinary 110% increase from a year earlier.
The results, released Tuesday after the market closed, offered another reminder of just how much money is flowing into AI infrastructure.
But a day earlier, Japanese robot manufacturer FANUC revealed a less obvious consequence of that spending.
The company has enough factory capacity to meet current demand. What it cannot reliably secure is every component needed to produce its equipment.
Memory chips, semiconductors, cables and mechanical parts are all becoming harder to obtain. Management warned that delivery delays could push customers toward competing suppliers.
The connection is important. AI infrastructure spending is generating new business for industrial equipment manufacturers, while simultaneously contributing to shortages of the components those manufacturers need.
And investors are becoming considerably less patient.
The selloff reaches the entire Physical AI industry
The Black Scarab Physical AI 15 declined 7.64% between the November 10 and November 17 closes, compared with losses of 2.40% for the S&P 500 and 2.94% for the Nasdaq Composite.
Every company in the index finished lower.
November 11–17
Weekly market performance
November 10–17 closes. Source: Black Scarab.
Ouster fell 20.3%, reversing a substantial portion of its previous week's 31.9% rally. Precision-gear manufacturer Nabtesco declined 17.8%, while automated material-handling specialist Daifuku lost 10.8%.
Siemens and Rockwell Automation held up comparatively well, falling 0.6% and 1.5%, respectively.
The selling coincided with a broader retreat from AI-related stocks, amid concerns about valuations, NVIDIA's leadership transition and the future pace of infrastructure investment.
NVIDIA's earnings arrived after Tuesday's close and therefore were not reflected in these returns.
The company's numbers suggest that demand for AI infrastructure remains exceptionally strong. But that doesn't settle the questions investors are asking about the businesses surrounding it.
For industrial suppliers, the challenge is increasingly about how much of that demand can be converted into revenue and, ultimately, profit.
Strong orders are creating a new set of problems
FANUC's management provided one of the clearest explanations this week.
In a Q&A published November 16, following its October 23 earnings briefing, the company identified component procurement as its principal production constraint.
It described the shortages as an industry-wide problem, partly driven by demand for semiconductor-manufacturing equipment. The company is working with additional suppliers, and even its engineering teams are helping address purchasing constraints.
There is a commercial risk beyond delayed revenue.
FANUC acknowledged that customers facing long delivery times might switch suppliers and that winning them back could be difficult.
FANUC
Where production gets held up
FANUC, October 23 briefing.
That's an unusually revealing statement from an established industrial manufacturer. Strong demand doesn't guarantee stronger earnings if fulfilling orders becomes more expensive or customers cannot wait.
Daifuku's latest results illustrate another part of the problem.
The Japanese company received ¥781.2 billion in orders during the first nine months of 2026, up 56.5% from a year earlier. Large cleanroom projects serving semiconductor production were a major contributor.
Revenue increased 13.9%, while operating profit rose 11%.
Daifuku raised its full-year operating-profit forecast from ¥113 billion to ¥119 billion, an encouraging sign. But its nine-month operating margin slipped from 15.5% to 15.1%.
Daifuku · January–September 2026
Orders, sales and profit growth
Japan GAAP, year over year. Source: Daifuku.
These results aren't contradictory. Orders represent work secured, while revenue and profit depend on when projects are executed, recognized and delivered.
Nabtesco, whose precision reduction gears are essential components in industrial robot joints, reported Component Solutions orders rising 23.6% and operating profit increasing 90.6%.
That looks like an exceptionally strong recovery.
But management also disclosed that some earlier profitability benefited from expenses arriving later than planned. Approximately ¥400 million of deferred Component Solutions costs are expected in the fourth quarter.
The company also acknowledged that material-cost increases had exceeded its original expectations. Full-year operating-profit guidance remained unchanged at ¥32.6 billion.
Investors are being reminded that a rising order book, improving margins and sustainable earnings growth are three different things.
Siemens shows what a more durable recovery can look like
Not every industrial supplier is struggling to convert stronger demand into profits.
Siemens delivered encouraging results on November 12.
Its Digital Industries division increased comparable quarterly revenue by 10%, while its reported profit margin climbed to 20.9% from 15.5%.
Part of that improvement reflected lower severance expenses.
But excluding severance charges, the margin still increased from 19.9% to 21.9%. That distinction matters because it demonstrates improvement beyond a favorable comparison with one-time costs.
Siemens Digital Industries · Fiscal Q4
Profit margin
Scale: 0–25%. Source: Siemens.
Siemens expects Digital Industries revenue to grow another 8–12% on a comparable basis in fiscal 2027, with a profit margin of 20–22%.
The company also reported record group free cash flow for fiscal 2026.
That combination of revenue growth, margin improvement and cash generation provides a useful benchmark for the rest of the industry.
It doesn't eliminate execution risk. But it offers stronger evidence of a sustainable operating recovery than order growth alone.
The broader factory economy tells a different story
The shortages affecting FANUC and other equipment suppliers might suggest that the entire manufacturing industry is running out of capacity.
The latest U.S. economic data says otherwise.
On November 17, the Federal Reserve reported that manufacturing output increased just 0.3% in October. Excluding motor vehicles and parts, factory production actually declined 0.1%.
Manufacturing capacity utilization reached only 76.0%, still below its long-term average.
In other words, the industrial economy has unused capacity even as certain suppliers struggle to satisfy demand.
AI infrastructure, semiconductor equipment and selected advanced-manufacturing markets are expanding rapidly. The recovery is much less uniform elsewhere.
Inflation adds another complication.
Friday's producer-price report showed U.S. final-demand prices rising 6.5% over twelve months, the largest annual increase since November 2022.
For automation projects, higher input costs and uncertain financing conditions can change the economics of an investment before the equipment is ever installed.
A manufacturer considering a robotic production line still needs to estimate its purchase price, installation costs, operating savings and payback period.
If those costs keep rising, some projects become harder to justify even when the underlying technology works.
The opportunity is shifting toward execution
Two developments this week offer an interesting perspective on where Physical AI companies may find opportunities.
Rockwell Automation announced the integration of NVIDIA Omniverse technologies into its Emulate3D industrial simulation software. It also introduced GPU-equipped industrial computing systems designed to bring more AI processing closer to production equipment.
The practical opportunity is to test factory designs, identify integration problems and run intelligent applications before or during deployment, rather than discovering every problem on a live production line.
These are technology announcements, not evidence of new revenue or proven customer savings. But they address costs that increasingly matter when manufacturers evaluate automation investments.
Another example came from Hadrian, which announced the acquisition of Pennsylvania's historic Kutztown Foundry & Machine.
As Black Scarab reported, Hadrian plans an additional $300 million investment in the site, combining existing casting expertise with software and automation to expand submarine-component production.
The market is different from semiconductor manufacturing, but the operating problem is familiar: downstream production cannot accelerate indefinitely if specialized components remain a bottleneck.
More sophisticated software and robotics may help. The economic payoff depends on whether they improve the quantity, quality and reliability of finished output.
What we're watching next
The next few weeks should provide a clearer picture of whether stronger industrial demand is translating into sustainable earnings.
For FANUC, the important development will be improved component availability and shorter delivery times. For Daifuku, it will be the pace at which major projects move from orders into revenue without further margin pressure. For Nabtesco, the fourth quarter will help establish how much of its margin recovery survives the return of deferred expenses.
Siemens has established a demanding benchmark for profitable growth. Rockwell's new industrial AI products will eventually need to demonstrate commercial adoption beyond their initial announcements.
The economic calendar also matters. The Federal Reserve's October meeting minutes are scheduled for November 18, followed by the October PCE inflation report on November 25.
Persistent inflation would make financing and project returns harder to manage. Easing price pressure would be welcome, although it wouldn't solve component shortages or execution delays.
Our outlook on Physical AI remains constructive, particularly where automation solves expensive, measurable industrial problems.
But the distinction between demand and profitable delivery is becoming more important.
NVIDIA's results show how much capital is still flowing into AI. The industrial companies supplying that expansion now need to show how efficiently they can convert the opportunity into earnings.
The next winners may not be the companies reporting the fastest order growth, but those capable of turning those orders into dependable shipments, sustainable margins and cash flow.
Sources
November 11–17, 2026. Returns use November 10–17 closes, excluding dividends and FX. FANUC’s Q&A covers its October 23 call.
The AI Boom Is Creating Its Own Bottlenecks
In this edition +
Explore the Physical AI 15 ↗$107.6 billion.
That's how much revenue NVIDIA generated from its Data Center business in a single quarter, an extraordinary 110% increase from a year earlier.
The results, released Tuesday after the market closed, offered another reminder of just how much money is flowing into AI infrastructure.
But a day earlier, Japanese robot manufacturer FANUC revealed a less obvious consequence of that spending.
The company has enough factory capacity to meet current demand. What it cannot reliably secure is every component needed to produce its equipment.
Memory chips, semiconductors, cables and mechanical parts are all becoming harder to obtain. Management warned that delivery delays could push customers toward competing suppliers.
The connection is important. AI infrastructure spending is generating new business for industrial equipment manufacturers, while simultaneously contributing to shortages of the components those manufacturers need.
And investors are becoming considerably less patient.
The selloff reaches the entire Physical AI industry
The Black Scarab Physical AI 15 declined 7.64% between the November 10 and November 17 closes, compared with losses of 2.40% for the S&P 500 and 2.94% for the Nasdaq Composite.
Every company in the index finished lower.
Weekly market performance
Ouster fell 20.3%, reversing a substantial portion of its previous week's 31.9% rally. Precision-gear manufacturer Nabtesco declined 17.8%, while automated material-handling specialist Daifuku lost 10.8%.
Siemens and Rockwell Automation held up comparatively well, falling 0.6% and 1.5%, respectively.
The selling coincided with a broader retreat from AI-related stocks, amid concerns about valuations, NVIDIA's leadership transition and the future pace of infrastructure investment.
NVIDIA's earnings arrived after Tuesday's close and therefore were not reflected in these returns.
The company's numbers suggest that demand for AI infrastructure remains exceptionally strong. But that doesn't settle the questions investors are asking about the businesses surrounding it.
For industrial suppliers, the challenge is increasingly about how much of that demand can be converted into revenue and, ultimately, profit.
Strong orders are creating a new set of problems
FANUC's management provided one of the clearest explanations this week.
In a Q&A published November 16, following its October 23 earnings briefing, the company identified component procurement as its principal production constraint.
It described the shortages as an industry-wide problem, partly driven by demand for semiconductor-manufacturing equipment. The company is working with additional suppliers, and even its engineering teams are helping address purchasing constraints.
There is a commercial risk beyond delayed revenue.
FANUC acknowledged that customers facing long delivery times might switch suppliers and that winning them back could be difficult.
Where production gets held up
That's an unusually revealing statement from an established industrial manufacturer. Strong demand doesn't guarantee stronger earnings if fulfilling orders becomes more expensive or customers cannot wait.
Daifuku's latest results illustrate another part of the problem.
The Japanese company received ¥781.2 billion in orders during the first nine months of 2026, up 56.5% from a year earlier. Large cleanroom projects serving semiconductor production were a major contributor.
Revenue increased 13.9%, while operating profit rose 11%.
Daifuku raised its full-year operating-profit forecast from ¥113 billion to ¥119 billion, an encouraging sign. But its nine-month operating margin slipped from 15.5% to 15.1%.
Orders, sales and profit growth
These results aren't contradictory. Orders represent work secured, while revenue and profit depend on when projects are executed, recognized and delivered.
Nabtesco, whose precision reduction gears are essential components in industrial robot joints, reported Component Solutions orders rising 23.6% and operating profit increasing 90.6%.
That looks like an exceptionally strong recovery.
But management also disclosed that some earlier profitability benefited from expenses arriving later than planned. Approximately ¥400 million of deferred Component Solutions costs are expected in the fourth quarter.
The company also acknowledged that material-cost increases had exceeded its original expectations. Full-year operating-profit guidance remained unchanged at ¥32.6 billion.
Investors are being reminded that a rising order book, improving margins and sustainable earnings growth are three different things.
Siemens shows what a more durable recovery can look like
Not every industrial supplier is struggling to convert stronger demand into profits.
Siemens delivered encouraging results on November 12.
Its Digital Industries division increased comparable quarterly revenue by 10%, while its reported profit margin climbed to 20.9% from 15.5%.
Part of that improvement reflected lower severance expenses.
But excluding severance charges, the margin still increased from 19.9% to 21.9%. That distinction matters because it demonstrates improvement beyond a favorable comparison with one-time costs.
Profit margin
Siemens expects Digital Industries revenue to grow another 8–12% on a comparable basis in fiscal 2027, with a profit margin of 20–22%.
The company also reported record group free cash flow for fiscal 2026.
That combination of revenue growth, margin improvement and cash generation provides a useful benchmark for the rest of the industry.
It doesn't eliminate execution risk. But it offers stronger evidence of a sustainable operating recovery than order growth alone.
The broader factory economy tells a different story
The shortages affecting FANUC and other equipment suppliers might suggest that the entire manufacturing industry is running out of capacity.
The latest U.S. economic data says otherwise.
On November 17, the Federal Reserve reported that manufacturing output increased just 0.3% in October. Excluding motor vehicles and parts, factory production actually declined 0.1%.
Manufacturing capacity utilization reached only 76.0%, still below its long-term average.
In other words, the industrial economy has unused capacity even as certain suppliers struggle to satisfy demand.
AI infrastructure, semiconductor equipment and selected advanced-manufacturing markets are expanding rapidly. The recovery is much less uniform elsewhere.
Inflation adds another complication.
Friday's producer-price report showed U.S. final-demand prices rising 6.5% over twelve months, the largest annual increase since November 2022.
For automation projects, higher input costs and uncertain financing conditions can change the economics of an investment before the equipment is ever installed.
A manufacturer considering a robotic production line still needs to estimate its purchase price, installation costs, operating savings and payback period.
If those costs keep rising, some projects become harder to justify even when the underlying technology works.
The opportunity is shifting toward execution
Two developments this week offer an interesting perspective on where Physical AI companies may find opportunities.
Rockwell Automation announced the integration of NVIDIA Omniverse technologies into its Emulate3D industrial simulation software. It also introduced GPU-equipped industrial computing systems designed to bring more AI processing closer to production equipment.
The practical opportunity is to test factory designs, identify integration problems and run intelligent applications before or during deployment, rather than discovering every problem on a live production line.
These are technology announcements, not evidence of new revenue or proven customer savings. But they address costs that increasingly matter when manufacturers evaluate automation investments.
Another example came from Hadrian, which announced the acquisition of Pennsylvania's historic Kutztown Foundry & Machine.
As Black Scarab reported, Hadrian plans an additional $300 million investment in the site, combining existing casting expertise with software and automation to expand submarine-component production.
The market is different from semiconductor manufacturing, but the operating problem is familiar: downstream production cannot accelerate indefinitely if specialized components remain a bottleneck.
More sophisticated software and robotics may help. The economic payoff depends on whether they improve the quantity, quality and reliability of finished output.
What we're watching next
The next few weeks should provide a clearer picture of whether stronger industrial demand is translating into sustainable earnings.
For FANUC, the important development will be improved component availability and shorter delivery times. For Daifuku, it will be the pace at which major projects move from orders into revenue without further margin pressure. For Nabtesco, the fourth quarter will help establish how much of its margin recovery survives the return of deferred expenses.
Siemens has established a demanding benchmark for profitable growth. Rockwell's new industrial AI products will eventually need to demonstrate commercial adoption beyond their initial announcements.
The economic calendar also matters. The Federal Reserve's October meeting minutes are scheduled for November 18, followed by the October PCE inflation report on November 25.
Persistent inflation would make financing and project returns harder to manage. Easing price pressure would be welcome, although it wouldn't solve component shortages or execution delays.
Our outlook on Physical AI remains constructive, particularly where automation solves expensive, measurable industrial problems.
But the distinction between demand and profitable delivery is becoming more important.
NVIDIA's results show how much capital is still flowing into AI. The industrial companies supplying that expansion now need to show how efficiently they can convert the opportunity into earnings.
The next winners may not be the companies reporting the fastest order growth, but those capable of turning those orders into dependable shipments, sustainable margins and cash flow.
Sources
November 11–17, 2026. Returns use November 10–17 closes, excluding dividends and FX. FANUC’s Q&A covers its October 23 call.
About the author
Rodolfo Garcia Calderoni, CFAFounder, Black Scarab