Sponsored by

Welcome to the iPrompt Newsletter
Friday Edition

A CRM so smart, it updates itself.

HubSpot's CRM is so smart, it now updates itself. Calls get logged and summarized the moment they end. New leads get researched and a first outreach drafted, automatically. Deals move forward on their own, with next steps flagged before you have to ask. 

Even the one repetitive task you've been meaning to fix can now run itself, no code required. Customers using it are already seeing the difference. 

This isn't the CRM you have to work hard to figure out. It's the one already working before you log in.

iPrompt Signals

AI & robotics investing, explained so you can actually act on it.

ISSUE 35 · FRIDAY 30 OCTOBER 2026

This was the week the biggest AI spenders leaned harder on outside money, and the terms they got split apart. Meta priced $25bn of stock. Alphabet disclosed $12.4bn of share sales and $29bn of debt for its quarter. Microsoft, still generating cash, sold $25bn of bonds at the tightest spread in the group. For owners, the question is shifting from how much each spends to what each gives up to pay for it.

Last Friday I expected reserves and debt first, shares later if at all. Meta went to shareholders while holding $82.45bn of cash. Nothing forced it. It chose to, and my sequencing call was wrong.

MARKET SNAPSHOT

One-day moves on the dates shown unless stated.

Session

Move

Driver

Tue 27 Oct

Meta −1.6% to $620.19

Equity raise priced the previous evening

Wed 28 Oct

S&P 500 +0.3%; Nasdaq +0.3%; SK Hynix about +4%

Fed holds at 3.75–4.00%; SK Hynix record quarter

Thu 29 Oct

Alphabet −12.6% to $283.82; Micron −5.0%; CoreWeave −5.0%; Microsoft −0.2%

Alphabet missed forecasts; second cash-negative quarter

Thu 29 Oct

S&P 500 −0.3%; Nasdaq −1.4%; chip index −1.4%; VIX 17.20; 10-year 4.98%

Yields fell 7bp as tech sold off

Thu 29 Oct, after hours

Amazon about −7%; Intel about −23%

Amazon's trailing free cash flow −$42.4bn; Intel's weak guide

VIX, the 'fear gauge', prices expected S&P 500 swings over 30 days; above 25 is anxiety. Week to Thursday: S&P 500 +0.7%, Nasdaq +0.7%, chip index +7.3%.

CHART OF THE WEEK

Read the status labels before the bar lengths. Built for a phone screen, so share it.

TOP HEADLINES

1 Meta priced $25bn of stock it will issue later

Launched on Monday at $22.5bn alongside $12.5bn of mandatory convertible preferred, the stock was upsized to $25bn at pricing that evening: 40m shares at $625, about five times covered. Banks sold borrowed shares; Meta issues the new ones and collects about $24.9bn at settlement, by 19 November 2027. The preferred pays 6.50% and converts in August 2029 into 17m to 20m shares. No new shares exist yet. Settled in full, the two add about 2.2% to 2.3% to today's count. A $60bn Class C programme is planned, unsigned.

NEW TO INVESTING

Two less familiar ways to sell shares. A forward sale fixes the price today; the company issues the shares and collects the cash later. A mandatory convertible pays a fixed dividend, then turns into ordinary shares on a set date. How many depends on the share price then, within a range set today.

2 Alphabet's second negative quarter came with shares already on sale

Its first revenue and earnings miss in nearly four years, despite Cloud growth of 103%. Free cash flow was −$9.07bn: $42.29bn from operations against $51.36bn of capex. It sold $12.4bn of stock through its $40bn programme, raised about $29bn of debt, bought back nothing and expects 2027 capex to 'increase significantly'. The shares fell 12.6%.

Correction to Issue 34: we described Alphabet's wallets as reserves, then bonds. It had already raised $44.75bn of equity in June. We missed it.

3 Microsoft borrowed at the tightest spread in the group

Its first high-grade bond sale since 2017: $25bn in ten tranches, coupons from 4.85% (2028) to 6.35% (2066), about $115bn of orders. The 2066 notes priced 1.05 points over Treasuries; 40-year bonds from Alphabet, Meta and Oracle trade 1.3 to 1.4 points over. Five-year credit protection costs 0.54 points a year on Microsoft, 1.16 on Meta, 2.75 on Oracle.

AI INFRASTRUCTURE AND GLOBAL

4 Anthropic's prospectus: fixed bills, flexible customers

Filed on Monday for a Nasdaq listing as ANT: first-half revenue $16.3bn, an operating loss of $693m, $131.7bn of cash and short-term investments. The compute total is left blank, but reported minimums include at least $114.5bn owed to Google, $110bn to Amazon and $44.9bn to Microsoft, used or not. Only $10.8bn of customer contracts carry fixed minimums.

5 SK Hynix's order book lengthened

Record Q3: ₩83.3tn of operating profit on ₩104.6tn of revenue. Long-term agreements cover over half of 2027–28 capacity across 14 major customers. They set volumes, not prepayments, so SK Hynix still funds its own build, with investment rising 'significantly' next year. Of its ₩100tn 2026 return target, this quarter's special dividend came from Kioxia gains, not operations.

TRIPWIRE TRACKER

'Happened' records the event. 'Thesis' records whether it moved the argument.

Tripwire (Issue 34)

Happened?

Thesis

Alphabet Q3: second negative quarter, higher 2027 plan

Yes. −$9.07bn; 2027 to rise 'significantly'

Moved. Debt and shareholders both carried more

Meta: first share sale for cash before Q4

Yes. Class A, not C; priced 26 Oct, issued at settlement

Moved. Our sequencing was wrong

A big buyer lowers its 2026 or 2027 capex

No. Alphabet up in 2027; Amazon holds about $220bn

Supplier half intact. Microsoft's July 'cut' was accounting

Anthropic S-1: what is fixed, and to whom

Partly. Filed; total left blank

Scored in the companion

Ibiden interim: Nvidia advance visible

Partly. Full-year operating profit guide to ¥133bn

Open. Advance booking unconfirmed

Fed, 28 Oct

Hold, 10–2; two votes for a hike

December hike priced near 85%

OUR INVESTING ANGLE

Everyone's watching how much the giants raise. I'd watch the terms.

The thesis: funding terms are now a measurable difference between the AI spenders, and I expect it to widen over the next two quarters. That's a claim about financial strength, not a stock call; valuation, earnings revisions and returns on the spending still drive the shares. But a dearer funder needs a higher return on the same build to leave owners better off.

How I'll judge it, for Microsoft, Alphabet, Meta, Amazon and Oracle:

The direct test: credit. The five-year CDS gap between Meta and Microsoft was 0.62 points on Thursday. Wider by 31 March 2027 supports the thesis; below 0.40 counts against it.

Supporting, not decisive: ownership. Track three things separately: dilution already in reported earnings per share, shares actually issued (Meta's forward settles by November 2027) and future conversion (its preferred, August 2029). A Q4 share count shows only the first.

Supporting, not decisive: returns. If the cheapest funders lag over the same period, that may mean the advantage was already in the price, not that it doesn't matter. Read returns alongside starting valuations and earnings revisions.

Better placed if I'm right: Microsoft, and suppliers with contracted order books such as SK Hynix. More pressured: Meta, Oracle and CoreWeave, whose 2031 notes yielded 15.58% on 20 October.

Where I could be wrong: spreads strip out the level of rates, but they can tighten for everyone when investors are eager, and then the differences matter less. Microsoft's 40-year notes still pay 6.35%: cheap relative to peers is not cheap in absolute terms. And equity is cheap when shares are richly valued; whether Meta's choice costs owners depends on what the projects return.

The companion, The Other Side of the Backlog, reads Anthropic's prospectus from the clouds' side and tests how much of a backlog to bank.

WHAT COULD GO WRONG WITH OUR ANGLE

1 Investors keep queuing. Meta's stock drew five times demand, Microsoft's bonds more than four. While that lasts, terms stay close.

2 The edge is priced. Microsoft slipped 0.2% on Thursday as Alphabet fell 12.6%. The strongest funder can still lose money at the wrong price.

3 Demand decides first. If cash-burning customers such as Anthropic slow their spending, every backlog shrinks, whatever the bond spread.

Size positions for the possibility that the thesis takes longer to prove, or proves wrong.

THREE IDEAS TO RESEARCH THIS WEEKEND

Not recommendations. Starting points for your own research.

1 Microsoft: how much of the backlog is one customer?

Why now: the cheapest funding in the group. The case: a funding edge matters most behind diversified demand. The risk: a July estimate put OpenAI at about a third of the $678bn commercial backlog. Tripwire: the December-quarter report, expected late January. Backlog growth excluding OpenAI below 20% means enterprise demand is slowing beneath the headline; 25% or more, July's pace, means it isn't. How to research: the 10-Q backlog note.

2 SK Hynix: how firm are those contracts?

I keep coming back to this. The agreements fix volumes, but penalties for walking away aren't disclosed. Why now: a record quarter and a longer order book. The risk: memory is cyclical, and management says it will flex spending to demand. Tripwire: Q4 results, expected late January. Coverage of 2027–28 below half, or a cut to 2027 investment, means the queue is shortening; coverage reaching into 2029 means it isn't. How to research: 000660.KS agreement disclosures, quarter by quarter.

3 Alphabet: bargain, or deposit?

Honestly, I don't know yet. The price: about $3.5tn, roughly 177 times trailing free cash flow, up from 78 times a week ago. The open question: how much of the $560.8bn cloud backlog is Anthropic. Tripwire: the Q4 report and annual filing, expected early February. Programme capacity below $27.6bn means more shares were sold; compare their average price with Thursday's $283.82. Sales below it mean owners funded more of the build at a lower price. How to research: the equity note and any named backlog customers.

FRAMEWORK: CHANGES THIS WEEK

Cloud stays 'funding led', now tracked by credit spread and share count. Meta moves to 'diluting at settlement'; Alphabet to 'diluting'. Microsoft is marked 'tightest spreads'. Not added: Anthropic, until it prices. Full seven-layer framework: [FRAMEWORK LINK]

WHAT WE ARE WATCHING

When

Checkpoint

Question

Week of 9 Nov (reported)

Anthropic IPO marketing

About $75bn expected. How much meets compute bills?

17 Nov · after close

Nvidia Q3

More equity in customers, after Meta?

9 Dec

Fed

A hike, priced near 85%?

Q1 2027

Meta Class C programme; distribution after 19 Jan

How fast it sells, and at what price

YOUR MOVE

Pick one AI holding. Find its newest bond's spread over Treasuries at a maturity close to Microsoft's 2066 notes (1.05 points), and its split-adjusted share count over eight quarters. Wider and rising mean its build costs owners more; whether that's worth it depends on what the build returns.

SHORT TAKE

Cloud and data-centre borrowers have raised more than $313bn of investment-grade bonds this year. If you own a broad corporate bond fund such as LQD, check its largest issuers.

Not a recommendation, a starting point.

Stay curious, and stay qualified.

R. Lauritsen

Know someone who owns Meta or Alphabet? Forward this before the Q4 reports.

P.S. Last week's poll: [POLL RESULT: X% spender, Y% supplier].

QUICK GLOSSARY

At-the-market programme Shares sold gradually into the market at current prices.

Backlog Contracted revenue not yet received (remaining performance obligations).

Credit default swap Insurance against default. Its yearly price, in points, is a quick read on credit risk.

Credit spread Extra yield over a government bond of similar maturity; it strips out the level of rates.

Forward sale Shares priced today, issued and paid for later.

Mandatory convertible preferred Pays a fixed dividend, then must turn into ordinary shares on a set date.

SOURCE AND METHOD NOTE

Data to the 29 October close; after-hours moves are early-evening readings. Meta's dilution (on about 2.57bn diluted shares, before January's Class C distribution), Alphabet's multiple and the CDS gap are our calculations.

Disclaimer: For information and education only, not financial advice. iPrompt Signals is not a registered investment adviser. Conduct your own research and consult a qualified financial professional before making investment decisions.

iPrompt Signals · Issue 35 · 30 October 2026 ·

Most AI Tools Weren't Built for This

B2B customer issues move across teams and systems, not through a single chat window. A new Harvard Business Review Analytic Services briefing paper, sponsored by Front, breaks down where AI tools fall short in B2B service and what to ask before you invest.