Morning Watchlist: Thursday Edition
A quick note from Behind the Markets
Wall Street keeps reducing the AI boom to a chip trade. That's lazy.
The real economy is repricing generators, bearings, cables, factories, and labor around the data-center buildout. Europe is discovering that defense autonomy requires industrial scale, not clever engineering. Regulators just drew a map for crypto capital formation. And an Australian software company found out what happens when your product becomes too essential to an industry.
Four stories, one question underneath each: what does the headline create, and what does it quietly break?
1) The AI Boom Is Hitting the Factory Floor
Generac is spending $250 million expanding factories that build large generators for data centers, and plans to add roughly 1,000 workers. Demand is radiating outward through cooling, transformers, construction equipment, wire, pipe, cement, and engineered bearings.
This is the part Wall Street misses. A data center is a giant industrial project with power-quality requirements, hard delivery deadlines, and thousands of suppliers. The most revealing detail in the reporting: manufacturers are going back to customers holding old purchase orders and asking for price increases simply to preserve delivery schedules. That's not a pricing-power theory. That's pricing power being exercised in the open.
The Stock: Generac Holdings (NYSE: GNRC)
Generac makes backup and prime power generation equipment, and it has become one of the primary suppliers of large-megawatt backup generators to hyperscale data centers. Its second quarter, reported July 29 (event risk cleared), showed the transformation in progress.
Commercial and industrial sales rose 29% to $556 million, driving total revenue up about 11% to $1.17 billion, with data-center products contributing more than $100 million in the quarter. Non-GAAP earnings of $2.91 beat consensus by nearly 45%. Adjusted EBITDA climbed to $291 million, or 24.8% of sales, from 17.7% a year earlier. The company secured a global supply agreement with a second hyperscale customer during the quarter, took roughly $1 billion of new orders in 90 days, and now carries a $1.6 billion data-center backlog — a figure that doesn't yet include any committed volume from that second hyperscaler. Management raised its C&I growth outlook to the low-30s percent range and called the opportunity "generational."
Now the three things you need to weigh against that, and keep in mind this warning applies directly to its anchor company.
First — a large chunk of the margin came from tariff refunds. Generac's gross margin expanded to 44.5% from 39.3%, and the company states that roughly $71 million in tariff refunds contributed about 6 percentage points of that gain. Strip it out and gross margin was roughly below the prior year. The earnings beat is real; the margin expansion is substantially a one-time item. (Fortune Brands, Home Depot, and PepsiCo have all leaned on tariff refunds in the same way this month — it's becoming a genuine pattern in this earnings season, and readers should look for it everywhere.)
Second, read the backlog's calendar. Of that $1.6 billion, about $250 million ships in the second half of 2026 and roughly $1.35 billion is scheduled for 2027 delivery. Management confirmed the backlog contains nothing for 2028 yet. That's excellent visibility into next year and no visibility beyond it — while the company is simultaneously tripling packaging and assembly capacity over the next year. Generac has firm orders through 2027 and is building for a 2028 it hasn't sold yet.
Third, the other half of the business is soft. Residential sales fell 2% to $621 million on a below-normal outage environment, and management cut its full-year residential growth outlook to the high single digits. Roughly half of Generac's revenue still depends on storms.
Bottom line: AI is an industrial demand multiplier. Follow suppliers with scarce capacity and cash discipline — and read the margin line to see how much of it is tariffs rather than pricing power.
2) Europe's Space Industry Has a Scale Problem
Italy's fragmented space sector is under pressure to consolidate as defense and government customers increasingly favor suppliers who can deliver integrated systems rather than excellent subsystems. Italy has world-class niche engineering firms; many lack the balance sheet and production scale for complex programs. Customers now want hardware, software, industrial capability, and cybersecurity in a single package.
The less obvious winners are the specialists who become indispensable to a consortium — optical communications, secure navigation, sensors, ground systems, testing. Recent contracts for Officina Stellare, including optical communications terminals for Leonardo and an Earth-observation program for the Italian Space Agency, show what happens when a niche technology proves it can deliver at industrial scale rather than in a laboratory.
The genuinely useful diligence questions here transfer to any government-adjacent supplier: is the company paid for development or for delivered systems? How concentrated is the customer base? How much working capital does it consume between milestone payments? Governments can announce strategic autonomy in a morning. Suppliers still have to make payroll while procurement grinds through committee.
Consolidation can unlock value, but it can also destroy the flexibility that made the small firm useful in the first place. A specialist absorbed into a prime often stops being the fast, focused supplier that won the contract.
Bottom line: Europe's space opportunity is shifting from clever components to integrated capacity. The winners will prove they can deliver a mission, not merely invent a subsystem.
3) The New Risk in Vertical SaaS Is Regulatory Concentration
Australia's competition regulator executed a search warrant at WiseTech Global seeking documents related to global logistics software and services. Shares fell roughly 12%. The company said it would cooperate.
The lesson travels far beyond one Australian software company, and it's worth internalizing because it inverts something most investors treat as unambiguously good.
Vertical SaaS businesses win by becoming the operating system for an industry. Customers build workflows around the platform, data lives inside it, and switching means re-engineering the business. That produces the sticky revenue and high margins investors pay premium multiples for. But the same integration that makes churn impossible can make a company a regulatory target — precisely because customers can't leave. Indispensability is the moat and the liability, and they're the same fact viewed from two directions.
So add an antitrust checklist to your SaaS checklist:
How dependent are customers, really? Not "would they be annoyed" — could they operate next quarter without it?
Can data move to a rival? Portability is the difference between a moat and a trap, and regulators look at exactly this.
Does the company expand through acquisition into adjacent workflows? Serial bolt-ons in the same vertical are what draws attention.
Are price increases supported by new value, or by the absence of alternatives? Ask what the pricing power is for.
A recurring-revenue multiple with no regulatory discount isn't conservative underwriting — it's an assumption that the moat will never be examined.
Bottom line: A vertical SaaS moat can attract regulators before it attracts a buyer. Competition and legal risk belong in the same model as churn.
4) The SEC Just Changed Crypto's Capital-Raising Math
On Tuesday the SEC proposed "Regulation Crypto Assets," and the specifics matter more than the headline.
The proposal creates two exemptions from Securities Act registration: a startup exemption allowing up to $5 million raised over four years, and a fundraising exemption allowing up to $75 million in any 12-month period. Issuers using either would provide principles-based narrative disclosures and would remain subject to antifraud and antimanipulation provisions. It also creates a conditional safe harbor that would apply once an issuer has completed or permanently ceased all the essential managerial efforts it promised — meaning a token could exit securities status once the network genuinely no longer depends on its sponsor.
Two caveats the excitement will skip. This is a proposal, not a rule — comments run for 60 days after Federal Register publication. And Commissioner Hester Peirce, among the SEC's most crypto-friendly voices, said plainly that the exemptions and safe harbor "will not fit every model." This follows the agency's March interpretation and runs alongside the CLARITY Act, which is still waiting on a September Senate vote.
The Stock: Coinbase Global (NASDAQ: COIN)
Coinbase is the most complete expression of the infrastructure thesis — exchange, custody, staking, stablecoin economics, and institutional services. Its strategic position is genuinely strong: crypto trading volume market share hit an all-time high of 10.3%, the third consecutive record quarter, achieved while total spot volume across the market fell 25% and crypto market capitalization dropped 11%. It is taking share in a shrinking market. The balance sheet is formidable — $8.6 billion of cash, roughly $10 billion of available resources, more than $2 billion returned to shareholders with another $2 billion authorized. Subscription and services revenue now represents 48% of net revenue, up from 29% less than two years ago, and adjusted EBITDA of $208 million marked the fourteenth consecutive positive quarter. Prediction-market revenue more than doubled sequentially.
And now the honest part, because Coinbase's own second quarter is the best available test of "infrastructure gets paid either way" — and it failed.
In the second quarter it came in at $555 million against roughly $599 million expected, declined sequentially from $584 million, and landed below the company's own guidance. Total revenue fell 19% year over year to $1.22 billion, missing consensus. The company posted a GAAP net loss of $359 million — its third consecutive quarterly loss — and cut headcount 14% to 4,321 employees.
Why? Because USDC interest income depends on interest rates and stablecoin balances, staking rewards depend on asset prices, and custody fees depend on the value of what's in custody. This revenue is less volatile than trading. It is not independent of the cycle. Third-quarter guidance for the segment is $500–580 million, implying it may decline again.
So own the framing correctly: Coinbase is the best-positioned infrastructure business in the sector, and it is still a leveraged bet on crypto activity. The stock carries a beta of 3.35 and fell about 58% over the past year. Nine analysts have cut estimates. Robinhood is launching a competing blockchain. Clearer rules should expand the addressable market over time — but a friendlier rulebook is not a floor under the share price.
Bottom line: Clearer crypto rules can expand the market while separating infrastructure from promotional issuance. Follow the settlement rails before chasing the next token — and don't mistake "less cyclical" for "not cyclical."
Before You Go
The market rewards the obvious beneficiary. The better trade is usually one layer deeper — but "one layer deeper" doesn't mean "safer," and today's issue makes that point four different ways.
AI creates a factory order — and the supplier filling it is tripling capacity for a year it hasn't sold yet, with margins flattered by a tariff refund. Rearmament creates a scale requirement — which may consolidate away the very specialists who made the sector interesting. A software moat creates regulatory exposure precisely because it works. And crypto clarity creates demand for trusted infrastructure — whose most successful operator just posted a loss while gaining record market share.
Identify what customers cannot operate without. Then ask who has the balance sheet, the qualification, and the patience to deliver it — and what happens to them if the thing everyone is extrapolating stops for four quarters.
Found this helpful? Share it with others.
Written by Behind the Markets