Your Week in 5: The AI Boom Gets Its First Real Invoice

This was the week the AI economy's money got audited in public. The Fed refused to make capital cheaper, Alphabet burned cash to build data centers, Amazon got docked for the same ambition, Reddit showed what AI answers cost the open web, and a prediction market raised a billion dollars betting that uncertainty itself is the product. One question ran through it all: who pays for this boom, and who gets paid?

1. The Fed holds at 3.75%–4% — and the next move may be up

• What’s happening: The Federal Reserve held its benchmark rate at 3.75%–4.00% on Wednesday in a 10–2 vote. Both dissenters — Governor Lisa Cook and Cleveland Fed President Beth Hammack — wanted a quarter-point hike, following September’s increase.
• Why it matters: Dissents now lean toward tightening, not easing. Futures markets put roughly 83% odds on a December hike. Capital is not getting cheaper for anyone reading this.
• The quote: Chair Kevin Warsh, days before the midterms, said “politics played no role in the decision” — and that he is “unwilling and unable to call” financial conditions tight.
• Zilck’s take: Build 2027 budgets assuming today’s cost of capital is the floor, not the ceiling. The Fed just told you the hurdle rate isn’t coming down to meet your pitch deck.

2. Alphabet goes cash-flow negative on AI. Microsoft shows the other path

• What’s happening: Alphabet grew revenue 25% to $127.6 billion, but free cash flow came in at negative $9.1 billion — a second straight negative quarter — as capex hit $51.3 billion. Shares sank 11%. Microsoft, reporting the same evening, grew Azure 48% and still generated $18.6 billion in free cash flow.
• Why it matters: Markets drew a hard line between AI spending that returns cash now and spending that promises to later. Alphabet funding the gap with $28.7 billion in new debt and $12.4 billion in share sales lands differently when rates are rising (see item 1).
• The quote: Microsoft CFO Amy Hood’s defense: “each dollar of CapEx has the potential to support more revenue and higher returns over time.” Sundar Pichai countered with scale: “a half-trillion dollar revenue run rate.”
• Zilck’s take: The AI trade didn’t crack — it matured. Cash conversion, not model benchmarks, is now the scoreboard. That discipline will trickle down to every startup’s next term sheet.

3. Amazon’s $202 billion quarter wasn’t enough

• What’s happening: Amazon posted Q3 revenue of $201.9 billion, up 12%, with AWS accelerating to 39% growth — yet the stock fell as EPS of $1.89 missed the $1.97 consensus and 2026 capex near $220 billion spooked investors.
• The e-commerce angle: Buried in the sell-off: Q4 guidance of $240–247 billion implies the biggest holiday quarter in retail history. Sellers and brands should plan inventory and ad spend to the guidance, not the stock chart.
• Zilck’s take: Amazon is two companies — a retail cash machine funding an AI land grab. If you sell on the platform, the land grab is your problem too: infrastructure costs have a way of becoming fee increases.

4. Kalshi raises $1 billion at $40 billion — up 82% in five months

• What’s happening: Prediction-market operator Kalshi raised $1 billion co-led by Sequoia and Wellington on Thursday — five months after a round that valued it at $22 billion.
• Why it matters: The week public markets punished AI capex, private capital paid a premium for market infrastructure: Kalshi is expanding past event contracts into CFTC-approved perpetual futures, positioning against CME and ICE.
• What to watch: Whether regulators keep blessing the blur between prediction markets, derivatives, and retail trading — the entire valuation rests on it.
• Zilck’s take: Founders, note the pattern: the fastest repricing in venture right now goes to companies that monetize volatility rather than fight it.

5. Reddit names the culprit for vanishing traffic: Google’s AI Mode

• What’s happening: On its Thursday earnings call, Reddit said search referrals fell by more than a million per week on average in October, with CEO Steve Huffman calling Google’s AI Overviews and AI Mode “the biggest” factor. Shares dropped about 18% after hours — despite Q4 revenue guidance of $1.0–1.03 billion, up 38–42%.
• The quote: Huffman, plainly: “The reality is the AI Mode does not drive traffic.” His adaptation plan: “We’ll likely be trading reach for engagement.”
• The marketing angle: If a top-ten web destination can’t hold its Google referrals, neither can your blog. Reddit’s hedge — over 80% of monetization comes from logged-in, in-app users — is the template: owned audience over borrowed traffic.
• Zilck’s take: Stop measuring content by sessions Google may never send. Measure it by the emails, accounts, and communities it converts. Reach is now rented; engagement is the asset.

The Narrative Thread

Every story this week is the same story: the AI boom’s invoice arrived, and it’s being routed to different departments. Alphabet and Amazon paid it on the balance sheet. The Fed — whose own statement cited “robust capital investment” in an economy running hot — is effectively charging interest on it. Reddit paid it in traffic, the open web’s new tax. And investors spent the week funding the toll booths: Kalshi’s $40 billion markets-for-everything bet, plus Re:Build’s $2.1 billion from General Catalyst and JPMorganChase to reindustrialize American hard tech — the week’s quietest giant deal, and a signal that “picks and shovels” now includes factories.

The takeaway for founders, marketers, and operators: run everything through one filter — how fast does this turn into cash? Price 2027 plans at today’s rates, not last year’s. Favor projects that pay back inside the window capital now costs. And move your growth math off borrowed traffic before Google finishes the job: the companies rewarded this week were the ones that own their distribution, their infrastructure, or their market. Rent as little of your business as possible.