Maurizio Pesce / Wikimedia Commons. Fremont factory, 2011. CC BY 2.0.

After a week of watching interest rates and worrying about demand, Tesla shareholders got something straightforward on Friday: 486,532 vehicles delivered in the third quarter, comfortably above the company-compiled analyst estimate of 461,974. Shares rose 4.65% to $370.59.

The surprise was welcome. The comparison with last year was less flattering. Deliveries fell 2.1% from 497,099, even though they improved from the second quarter. Tesla beat expectations; it hasn’t yet returned to year-over-year growth. Both can be true, and both matter when deciding how much of Friday’s rally to credit to an improving business.

Production came in at 464,391 vehicles, meaning deliveries exceeded output by 22,141. That is a better inventory direction than building cars faster than customers take them home. It does not tell us what discounts, financing offers or product mix were needed to complete those sales. The production and delivery breakdown is the cleanest starting point.

The Model 3 and Model Y accounted for 478,237 deliveries. Everything else contributed 8,295. For all the attention on new vehicles and autonomy, the core models still do almost all the delivery work. Energy storage deployments were 13.7 gigawatt-hours, another business to watch when the full financial results arrive.

The next question is what Tesla earned on those deliveries. More cars can help spread factory costs, but volume purchased through heavier incentives is less valuable than volume arriving at stable prices. The October 21 earnings report will tell us much more about that trade-off than Friday’s release did.

And the weekly result keeps the celebration in proportion: Tesla still finished 0.41% below the previous Friday. The delivery beat repaired most of the damage. It did not turn the week into a breakout.

Brandon Carson; crop by Another Believer / Wikimedia Commons. Nike campus, 2010. CC BY 2.0.

Nike’s Friday decline was smaller than its initial overnight selloff, but the final result was hardly reassuring: down 3.64% to $33.87, and 5.26% lower for the week. Investors had another quarter of evidence that fixing the business will take time.

Thursday’s report put fiscal first-quarter revenue at $11.2 billion, down 4%. Direct sales fell 8%, including a 13% decline online. Wholesale revenue slipped only 1%. Gross margin improved to 42.8%, helped by lower logistics and warehousing costs, while earnings were 48 cents a share.

There is progress in those numbers, just not the kind that settles the demand question. A cheaper distribution operation improves profit on each sale. It doesn’t necessarily persuade someone to buy another pair of shoes. The smaller wholesale decline suggests Nike’s retail partners are holding up better than its own channels, but that is a relative improvement inside a shrinking sales base.

The forecast was the harder part: a high-single-digit revenue decline for fiscal 2027, with adjusted earnings of $1.15–$1.35 a share. Reuters reported that China sales fell 26% on a currency-neutral basis. The reporting on China and further job cuts explains why this extends beyond one weak quarter.

Nike’s new Pace program targets $2.5 billion in cumulative savings through fiscal 2031, before charges and reinvestment. That is a multiyear total, not an annual profit boost. The company expects roughly $1 billion in pretax restructuring charges over the period. Nike’s release lays out the guidance and savings plan.

The investment argument now depends on the sequence. Cutting costs can protect the business while products and distribution improve. But if sales keep falling, some of those savings merely offset lost revenue. We’d learn more from a sustained improvement in full-price demand than another ambitious savings announcement.

Thingreenline4546 / Wikimedia Commons. Taichung, 2023. CC BY-SA 4.0.

Micron’s $54.23 billion quarter and $61.5 billion next-quarter revenue forecast were among the week’s most striking numbers. Yet the stock ended Friday at $1,074.89, down 0.68% for the week, after losing 2.05% in the final session.

The company is delivering extraordinary growth. Investors are deciding how long it can last—and how much they already paid for it.

Wednesday’s fiscal fourth-quarter release showed adjusted earnings of $33.42 a share and an adjusted gross margin of 87%. Quarterly sales alone exceeded the $37.38 billion Micron generated in the entire previous fiscal year. The new revenue outlook is $61.5 billion, plus or minus $1.5 billion. The results and forecast show how dramatically the business has changed.

Pricing explains much of it. Management said DRAM prices increased in the high teens from the preceding quarter, while shipments grew by a mid-single-digit percentage. NAND prices rose roughly 30%, against shipment growth of about 10%. Selling more helps; getting paid considerably more for the same amount of memory helps much more.

Customers are also making longer commitments. Micron disclosed roughly $150 billion in remaining performance obligations and $32 billion in customer financial commitments, mainly deposits. Those are different measures of future business and customer support. They should not be added together as another revenue figure. Management’s prepared remarks explain the contracts and capacity plans.

The expansion is expensive: first-half fiscal 2027 capital spending is projected around $25 billion, with more in the second half. That is why the stock debate has moved beyond whether AI needs memory. It plainly does. The harder question is whether today’s exceptional pricing survives as manufacturers build more capacity.

TechCrunch / Wikimedia Commons. Dario Amodei, 2023. CC BY 2.0.

AI customers need chips. Increasingly, they also need someone to help pay for them.

Bloomberg reported Friday that Broadcom’s banking group is assembling $60 billion in financing for chips and infrastructure benefiting Anthropic and other companies. The proposed structure includes $42 billion in senior-secured debt and $18 billion in junior debt, with Blackstone leading the latter. Broadcom rose 3.35% Friday, finishing the week up 0.66%.

The financing is still being arranged. It is not $60 billion of completed chip sales, and it should not be piled on top of every other Anthropic financing headline as wholly separate spending. Bloomberg’s breakdown of the proposed package distinguishes the lenders and debt layers.

For Broadcom, outside funding could convert customers’ computing ambitions into orders that can actually be delivered and paid for. For lenders, the calculation is different: will the equipment generate enough income, for long enough, to cover interest and principal?

That creates a useful distinction for shareholders. Strong demand for AI services would support the whole arrangement. Easier access to borrowing can accelerate construction before that demand has produced sufficient cash. Both can make chip orders look strong today, but they leave very different risks if customers’ growth disappoints. Watch the financing terms as closely as the headline amount.

Planet Labs PBC, via NPR / KPBS. Suncatcher prototype.

Google’s Project Suncatcher moved from an idea to an actual satellite this week. The prototype, built with Planet, launched Thursday aboard SpaceX’s Transporter-18 mission. Google said contact had been established and the satellite was operating as expected.

The immediate job is practical: test how Google’s Tensor Processing Units handle spaceflight, radiation and temperature extremes. It is an early experiment in whether machine-learning infrastructure could eventually operate in orbit. Google’s mission update and research describe what the prototype will measure.

For Alphabet shareholders, the distinction between a successful launch and a successful business matters. Getting equipment into orbit provides an opportunity to collect data. It does not establish the cost of running a useful service there, keeping it connected or replacing failed hardware.

There is room to find the engineering exciting without putting hypothetical space-computing revenue into next year’s forecast. The next meaningful update will be evidence about how the chips perform and what the tests reveal about reliability. That would make the larger proposal easier to assess. For now, this remains research spending with a potentially interesting payoff much further out.