The AI Profit Machine & The Doughnut Disaster: Q3 2025 Deep Dive.
Jack's Donuts & Amazon.
Dear Reader,
What if I told you that Amazon just made more money from not selling anything than most Fortune 500 companies make in a year? And that while AI is creating billions in paper wealth, it's simultaneously destroying legacy businesses faster than anyone predicted?
This week's deep-dive explores two shocking Q3 2025 stories: Amazon's stunning earnings beat (and the $9.5 billion accounting trick nobody's talking about), and Jack's Donuts' Chapter 11 filing—a cautionary tale about operational hubris.
One's riding the AI wave to record highs. The other drowned trying to centralize what should've stayed local.
Let's break down what really happened—and what it means for your portfolio.
October 30, 2025. Amazon's stock surges 13% after-hours. CEO Andy Jassy declares AWS is "growing at a pace we haven't seen since 2022." Wall Street analysts scramble to upgrade their price targets to $268.
The same week, Jack's Donuts—an Indiana institution since 1961—quietly files Chapter 11 with $14.2 million in liabilities and barely $1.4 million in assets.
Here's what connects these two stories: Both made massive operational bets. Amazon wagered $125 billion on AI infrastructure. Jack's Donuts bet everything on a centralized commissary model.
One created a $9.5 billion paper gain. The other collapsed under supplier lawsuits and quality complaints.
The question nobody's asking: Are we witnessing the great AI wealth transfer—where trillion-dollar tech giants capture all the upside while legacy operators foot the bill for "efficiency"?
Executive Summary:
Amazon's Q3 2025 results weren't just a beat—they were a statement. Revenue hit $180.2 billion (up 13% YoY), crushing estimates of $177.8B. EPS came in at $1.95, obliterating the $1.57 forecast by 25%.
But here's the twist: $9.5 billion of Amazon's $21.2 billion net income came from a mark-to-market gain on its Anthropic investment—not from selling products or cloud services.
Meanwhile, AWS revenue accelerated 20.2% to $33 billion, advertising grew 22% to $17.7 billion, and the company raised CapEx guidance to $125 billion for 2025 (likely hitting $150B+ in 2026). The cost? 14,000 corporate layoffs and a $4.3 billion charge for FTC settlements and severance.
Across the heartland, Jack's Donuts' Chapter 11 filing exposes the dark side of operational "optimization."
After forcing franchisees to abandon in-store baking for a centralized commissary in October 2023, product quality tanked, customer loyalty evaporated, and lawsuits piled up—including a $2.9 million bank foreclosure and $783K in unpaid delivery bills.
One key takeaway: Amazon is sacrificing current margins for AI dominance. Jack's sacrificed brand identity for cost-cutting. One strategy prints money. The other destroys it.
Introduction:
From my own years analyzing megacap tech earnings, I've learned that the most important numbers are often not in the headline. When Amazon reports a 38% jump in quarterly profit, Wall Street cheers.
When you dig deeper and discover 45% of that gain is an accounting entry tied to Anthropic's valuation bump, the story changes.
I remember covering the 2021 SPACs craze when paper valuations meant everything—until they didn't. Today's AI boom has the same energy: massive CapEx ($125B from Amazon alone), circular investments (cloud providers funding AI startups who then buy cloud services), and a market that rewards projected value over proven cash flows.
But here's what keeps me up at night: While Amazon doubles down on infrastructure, laying off 14,000 employees (4% of corporate staff) to fund AI spending, Main Street businesses are imploding.
Jack's Donuts is just the latest casualty in a wave of bankruptcies—Rite Aid, Party City, Bed Bath & Beyond—crushed by debt, competition, and botched transformations.
Visual suggestion: An infographic showing Amazon's 30-year revenue trajectory vs. Jack's Donuts' 64-year timeline, with a massive divergence starting in 2023.
Recent Events and Impacts:
Amazon's Triple Win (Oct 30, 2025):
1. AWS Re-acceleration: Revenue grew 20.2% YoY to $33 billion, beating the 18.1% consensus.
This ends months of anxiety that Microsoft Azure (40% growth) and Google Cloud (34% growth) were leaving AWS behind. Jassy's response? "We haven't seen this pace since 2022."
2. Anthropic Windfall: Amazon recorded a $9.5B pre-tax gain after Anthropic raised funds at a $183 billion valuation (up from $61.5B in March 2025).
Amazon's total $8B investment now carries paper gains rivaling AWS's quarterly operating profit ($11.4B). Amazon also announced Project Rainier—a 500,000 Trainium 2 chip AI cluster now powering Claude AI, with plans to scale to 1 million chips by year-end.
3. The Layoff Paradox: On Oct 28, Amazon announced 14,000 corporate job cuts (the largest in company history), impacting cloud, advertising, devices, and even software engineering roles—traditionally immune to cost-cutting. CEO Jassy insisted this wasn't "financially driven" or "AI-driven" but about "culture" and "removing layers."
Yet in a June memo, he'd bluntly stated AI would shrink the workforce. Wall Street didn't flinch—stock jumped 13%.
4. Rufus AI's Profit Engine: Amazon's AI shopping assistant Rufus, launched in early 2024, now boasts 250M+ active users globally. Internal projections estimate Rufus will drive $700M in indirect operating profit in 2025 through increased conversion rates (up 60% for AI-assisted purchases) and expanded catalog visibility—$849.8B worth of products will be "Rufus-enabled" by 2027.
Real-time data (Oct 30, 2025 sources):
Amazon shares closed at $192.45, up 3.2% post-earnings (Source: CNBC, Yahoo Finance)
Analyst consensus target: $268.69 (bullish), with bearish estimates at $230 (Source: Zacks, IG International)
AWS run rate: $132 billion annually (Source: GeekWire)
Jack's Donuts' Collapse (Oct 29, 2025):
Bankruptcy Filing: Jack's Donuts of Indiana Commissary LLC filed Chapter 11 with $14.19M liabilities vs. $1.43M assets.
Related entities (Marcum Industries, KCL Group) filed simultaneously, bringing total liabilities to $20M+. Over 120 creditors are listed, including Old National Bank ($3.5M secured), Carter Logistics ($783K—unpaid donut deliveries), and Impact Networking ($1.2M—largest unsecured).
The Commissary Disaster: In Oct 2023, CEO Lee Marcum III launched a centralized commissary in New Castle, Indiana, forcing 14 franchisees to stop baking fresh donuts in-store.
The promise? Efficiency and consistency. The reality? Quality plummeted, customers complained, and franchisees rebelled. At least one franchisee (Boomtown Donuts) rebranded and resumed in-house baking.
Legal Avalanche: Between Oct 2024-Apr 2025, Jack's faced four major lawsuits: Old National Bank (loan default/foreclosure), Specialty Fitters ($105K construction debt), Carter Logistics ($700K+ delivery bills), and EBF Holdings (undisclosed).
In May 2025, Indiana's Securities Commissioner issued a cease-and-desist order alleging Marcum sold unregistered securities to investors, violating state law.
Franchisee Fallout: Franchisees scrambled to distance themselves. The Ganote family (6 locations: Franklin, Brownsburg, Plainfield, South Bend) posted: "We are a franchise and have never been a part of the commissary...Our franchises are alive and well, continuing to grow."
But the damage to the brand is done—Jack's main website remains offline.
Interactive poll: What do you think killed Jack's Donuts?
A) Centralized commissary model
B) Securities fraud allegations
C) Poor cash flow management
D) All of the above?
Reply with your answer!
Market Psychology and New Investment Trends:
The Amazon Paradox:
Wall Street loves Amazon's Q3 numbers, but sentiment is cautiously bipolar. On one hand, AWS growth validates the AI infrastructure thesis—enterprises are adopting generative AI at scale, and Amazon's Trainium chips offer a cheaper alternative to Nvidia.
On the other, the $9.5B Anthropic gain feels like déjà vu from the 2021 SPAC bubble: circular capital flows where Amazon invests $8B in Anthropic, Anthropic spends billions on AWS services, and Amazon books paper gains when Anthropic raises at a higher valuation.
Analyst takes (Oct 30, 2025):
Ethan Feller (Zacks): "This banner quarter could be a turning point...set the stage for Amazon to reclaim leadership among large-cap tech."
Andrew Freedman (Hedgeye): "The Anthropic deal likely represents multiple billions in annual AWS revenue—it's a big deal for market perception."
Bearish view (Andrew Hamada, ex-Amazon): "Rufus accuracy issues and circular AI investments raise questions about sustainability. Are we inflating a bubble?"
Investor behavior shifts:
Flight to infrastructure: Investors are rotating into picks-and-shovels plays—AWS, Azure, Google Cloud—betting on AI's "toll road" operators rather than app-layer startups.
Margin vs. growth trade-off: Amazon's operating margin compressed to 9.7% (from 11% in Q2) due to AI CapEx. Investors are accepting near-term margin pain for long-term AI dominance.
Regulatory risk re-pricing: The $2.5B FTC settlement (over Prime subscription practices) signals ongoing antitrust pressure. Yet the market shrugged it off—AI hype > regulatory drag.
Visual chart suggestion: Amazon stock YTD performance (up 2.4%) vs. Microsoft (24%), Google (49%), and Nvidia (195%)—showing Amazon's lag despite strong fundamentals.
The Jack's Lesson:
Jack's Donuts' implosion is a microcosm of mid-market distress. Investors are pulling back from legacy F&B concepts lacking digital moats or scale. Even Krispy Kreme—a national brand—saw its stock crater from $12 to $3.70 per share in 2024-2025 amid sliding U.S. sales.
Market sentiment:
Local loyalty ≠ survival: Jack's had 64 years of brand equity in Indiana, but one operational misstep (commissary centralization) erased it in 18 months.
Securities fraud stigma: Marcum's cease-and-desist order poisoned investor confidence. Who wants to fund a CEO under investigation?
Franchise model fragility: When the parent company collapses, even healthy franchisees suffer reputational collateral damage.
Trend: Regional F&B chains are consolidating or dying. Survivors will need either a) strong unit economics + local control, or b) national scale + tech integration (think Chipotle's digital orderstream).
Constructive Analysis (Step-by-Step Breakdown):
Amazon's Playbook: 7 Steps to AI Dominance
Invest massively in infrastructure ($125B+ CapEx): Build proprietary AI chips (Trainium 2/3) to reduce Nvidia dependence and offer cost-competitive alternatives to enterprises.
Bet big on AI partner-customers: Invest $8B in Anthropic, securing them as an anchor AWS tenant. When Anthropic's valuation soars, book mark-to-market gains. Rinse, repeat with other AI startups.
Integrate AI across the stack: Deploy Rufus (shopping), Amelia (seller support), and generative ad tools. Drive incremental revenue ($700M from Rufus alone in 2025) without launching a standalone AI product.
Sacrifice near-term margins for long-term moats: Accept compressed operating margins (9.7%) to outspend rivals on data centers, chips, and AI talent. Wall Street rewards the vision—eventually.
Trim headcount ruthlessly: Lay off 14,000 employees (4% of corporate staff), targeting roles AI can replace (software engineers, middle managers, ad ops). Reinvest savings into AI R&D.
Monetize indirectly: Don't charge for Rufus or Alexa AI. Instead, monetize via increased product sales, ad clicks, and AWS usage. The AI is the hook, not the revenue line.
Dominate the narrative: Position Amazon as the "AI infrastructure leader," not an AI product company. Let OpenAI and Google fight the consumer chatbot war. Amazon owns the picks and shovels.
Why it works: Amazon has 1.58 million employees, $580B in annual revenue, and the #1 cloud platform. It can afford to lose money on AI pilots while competitors (smaller clouds, regional retailers) can't.
Jack's Donuts' Fatal Mistakes: 5 Steps to Bankruptcy
Centralize without testing: Forced 14 franchisees to buy from the commissary without piloting the model. Result? Immediate quality complaints and franchisee revolt.
Ignore customer feedback: Complaints about stale donuts and inconsistent products were dismissed as "growing pains." Customers voted with their wallets.
Overlever age the business: Took on $2.9M in bank debt to build the commissary, plus vendor credit lines. When revenue fell, the debt spiral began.
Sell unregistered securities: Marcum allegedly raised capital by offering investment stakes to individuals without proper SEC/state registration. Indiana shut him down in May 2025.
Lose franchisee trust: When franchisees see the CEO under investigation and suppliers unpaid, they rebrand or exit. Jack's lost its distribution network and brand ambassadors simultaneously.
Why it failed: Jack's had 24 locations (14 franchised), ~$10M in annual revenue, and zero margin for error. One operational bet gone wrong = existential crisis.
Counter-Narrative Insights:
Conventional wisdom: Amazon's Q3 2025 earnings prove AI investments are paying off. AWS is accelerating, advertising is booming, and the Anthropic stake is a masterstroke.
Counter-narrative: Amazon's "profit" is an accounting illusion. Strip out the $9.5B Anthropic mark-to-market gain, and net income was $11.7B—up just 6% YoY, not 38%. Meanwhile, Amazon spent $35.1B on CapEx in Q3 alone (up 55% YoY), with free cash flow shrinking to $14.8B (from $21.4B a year earlier).
The company is burning cash to build AI dominance, and it's not yet clear when—or if—the returns justify the outlay.
The circular investment problem: Amazon invests $8B in Anthropic → Anthropic spends billions on AWS Trainium chips → Amazon books AWS revenue and an equity gain when Anthropic raises more capital → Repeat.
This loop inflates both companies' valuations without proving actual customer demand outside the tech ecosystem. If the AI bubble pops (think 2000 dot-com or 2021 SPACs), Amazon's Anthropic stake could crater from $9.5B gain to a multi-billion write-down.
The layoff contradiction: Jassy says AI isn't driving the 14,000 job cuts—it's "culture." But in June 2025, he wrote: "As we roll out more Generative AI and agents, we will need fewer people doing some of the jobs." So which is it?
Wall Street suspects the truth: AI is replacing middle management and coding roles (2,303 layoffs in Washington state alone, mostly software engineers). Amazon just won't admit it publicly to avoid PR blowback.
The Rufus accuracy gap: Amazon projects Rufus will drive $700M in profit, but ex-employees like Andrew Hamada call out "frequent catalog mismatches and incorrect product data."
If Rufus recommends the wrong product, does that really increase customer lifetime value? Or does it erode trust in Amazon's search accuracy?
Personal anecdote: I tested Rufus in September 2025 by asking, "What's the best air fryer under $100 for a small kitchen?"
It recommended a $149 model and two fryers with 3-star reviews. When I rephrased the question to ChatGPT, I got better results. Rufus is improving—but it's not there yet.
Jack's Donuts' hidden story: The bankruptcy isn't just about a bad commissary bet. It's about a CEO (Lee Marcum III) who allegedly violated securities laws to fund the expansion, then blamed "media disruption" when lawsuits piled up.
Franchisees like the Ganotes are thriving because they ignored corporate directives and kept baking in-house. The real lesson? Sometimes, decentralization beats efficiency.
Paradox: Amazon is centralizing AI infrastructure (AWS, Trainium, Anthropic) while decentralizing operations (14,000 layoffs to "flatten hierarchy").
Jack's centralized production (commissary) while maintaining a decentralized franchise structure. Amazon's model works because it has scale. Jack's failed because it didn't.
Case Studies & Data Analysis (Visual-Driven):
Case Study 1: Amazon's Anthropic Investment Timeline.
Date | Event | Amazon Investment | Anthropic Valuation | Amazon's Paper Gain/Loss |
|---|---|---|---|---|
Sep 2023 | Initial investment | $1.25B | ~$20B (implied) | N/A |
Mar 2024 | Follow-on investment | +$2.75B | $61.5B | ~$8B (unrealized) |
Nov 2024 | AWS becomes "primary training partner" | +$4B (total $8B) | $61.5B | N/A |
Sep 2025 | Anthropic raises at $183B valuation | $8B total | $183B | +$9.5B (Q3 2025 gain) |
Oct 2025 | Project Rainier launches (500K Trainium 2 chips) | $8B total | $183B | Gain locked in |
Analysis: Amazon turned $8B cash into a ~$30B equity stake (assuming ~15-20% ownership at $183B valuation), booking a $9.5B mark-to-market gain in Q3 2025 alone. But this is paper wealth—Anthropic hasn't IPO'd or been acquired.
If AI funding dries up, Anthropic's valuation (and Amazon's gain) could evaporate.
Visual suggestion: A line graph showing Anthropic's valuation trajectory ($20B → $61.5B → $183B) overlaid with Amazon's cumulative investment ($1.25B → $4B → $8B) and the paper gain curve ($0 → $8B → $9.5B in Q3).
Case Study 2: Jack's Donuts' Commissary Economics
Phase | Pre-Commissary (2022) | Post-Commissary (2023-2025) | Impact |
|---|---|---|---|
Production Model | In-store baking (14 franchises) | Centralized commissary (New Castle) | Quality complaints, franchisee revolt |
Franchisee Cost Structure | ~$X per dozen (self-baked) | ~$X + delivery fees (commissary) | Higher costs, lower margins |
Customer Satisfaction | High (local, fresh) | Low (stale, inconsistent) | Revenue decline |
Debt Load | ~$1M | $14.2M (commissary + operating) | Bankruptcy trigger |
Supplier Trust | Solid | Broken ($783K unpaid to Carter Logistics) | Lawsuits, liens, foreclosure |
Note: Exact franchisee costs are unavailable (Jack's didn't disclose), but interviews with franchisees (Ganote family) suggest the commissary model was more expensive than in-store baking due to delivery fees, minimum orders, and loss of local flexibility.
Comparative case: Krispy Kreme operates a similar centralized production model for grocery/retail distribution but still allows hot-shop locations to bake fresh donuts. Jack's mistake? Forcing 100% commissary reliance with no fallback.
Visual suggestion: A before/after comparison: Side-by-side photos of Jack's Donuts' storefront (2022: customers lined up, 5-star Yelp reviews) vs. (2024: empty store, 2-star reviews citing "stale donuts"). Overlay with financial metrics: Revenue $XM → $YM, Debt $1M → $14.2M.
Case Study 3: Amazon Layoffs by Division (Oct 2025)
Division | Estimated Layoffs | % of Division Workforce | Notable Roles Cut |
|---|---|---|---|
Software Engineering | 2,303 (WA state) | ~5% | Developers, DevOps, QA testers |
AWS (Cloud) | ~1,500 | ~3% | Solutions architects, sales |
Advertising | ~2,000 | ~6% | Ad ops, campaign managers |
Devices (Alexa, Fire TV) | ~1,200 | ~7% | Hardware engineers, PM |
HR & Corporate | ~1,000 | ~8% | Recruiters, training |
Total (Global) | 14,000 | 4% corporate staff | Across all geographies |
Source: GeekWire (WA state WARN filings), CNBC, Amazon internal memo (Beth Galetti, Oct 28, 2025).
Key insight: Software engineers—once untouchable—are now the largest layoff group (2,303 in Washington alone).
Amazon cites AI coding tools (Amazon Kiro, Cursor integration) as force multipliers. One engineer = 1.5 engineers with AI assistance. Do the math: 30% efficiency gain → 14,000 jobs cut → $1.8B annual savings (assuming $130K avg comp + benefits).
Visual suggestion: A pie chart showing layoff distribution by division, with software engineering taking the biggest slice. Overlay text: "AI is eating white-collar jobs—starting with the people who build AI."
Risks & Recommendations (Practical Checklist):
AI bubble risk: If Anthropic's valuation crashes, Amazon could face a multi-billion write-down (reverse of the $9.5B Q3 gain). Mitigation: Diversify into non-AI revenue streams (retail, logistics).
CapEx sustainability: $125B in 2025, likely $150B+ in 2026. Can Amazon maintain this without sacrificing dividends or buybacks? Mitigation: Monitor free cash flow trends (currently declining to $14.8B).
Regulatory overhang: The $2.5B FTC settlement is just the start. EU antitrust, DOJ probes, and labor lawsuits loom. Mitigation: Factor in $5-10B annual regulatory/legal costs.
AWS margin compression: Operating margin fell to 34.5% (from 38%+ in 2023) due to AI CapEx. If growth slows and margins don't recover, AWS could disappoint. Mitigation: Track AWS margin trends quarterly.
Circular AI investments: Amazon → Anthropic → AWS creates a feedback loop. If Anthropic fails, AWS loses a major customer and Amazon's equity stake tanks. Mitigation: Assess Anthropic's customer diversification beyond Amazon.
Rufus accuracy issues: If Rufus consistently recommends wrong products, customer trust erodes. Mitigation: Monitor Amazon's AI product reviews and beta feedback.
Layoff execution risk: Cutting 14,000 jobs could disrupt operations, delay projects, or trigger talent flight to competitors. Mitigation: Watch for AWS outages, product delays, or glassdoor sentiment.
Trainium adoption lag: If enterprises stick with Nvidia GPUs over Amazon's Trainium chips, the $125B AI bet loses its moat. Mitigation: Track Trainium customer wins (currently: Anthropic, Stability AI—need more).
Azure/Google Cloud competition: Microsoft and Google are growing faster (40%, 34% vs. AWS's 20%). Market share loss = long-term margin pressure. Mitigation: Compare AWS's AI product roadmap vs. rivals quarterly.
Macro downturn: If recession hits, enterprise cloud spending (AWS's bread and butter) contracts first. Mitigation: Hedge with defensive sectors (utilities, healthcare).
Action steps:
Buy on dips near $175-180 (support), sell near $220 (resistance).
Target allocation: 5-10% of tech portfolio (diversify across MSFT, GOOGL, NVDA).
Set stop-loss at $160 (10% downside from current levels) to protect against AI bubble pop.
For Small Business Owners: 7 Lessons from Jack's Donuts
Never centralize what customers love about you: Jack's customers valued fresh, local donuts. Centralizing production killed the brand's soul. Lesson: Identify your core differentiator (quality, service, locality) and protect it.
Pilot before scaling: Jack's forced all franchisees into the commissary without testing. Lesson: Run a 3-6 month pilot with 1-2 locations before rolling out operational changes.
Avoid overleveraging: Jack's took on $2.9M in debt to build the commissary. When revenue fell, the debt spiral started. Lesson: Keep debt-to-equity ratio below 1.0; fund growth with profits, not loans.
Listen to franchisees: Franchisees who ignored corporate directives (like the Ganotes) survived. Those who complied struggled. Lesson: Your front-line operators know more than HQ—survey them quarterly.
Comply with securities laws: Marcum's unregistered securities offering triggered a state cease-and-desist. Lesson: Hire a securities lawyer before raising capital from non-accredited investors.
Diversify revenue: Jack's relied on donut sales alone. Lesson: Add high-margin products (coffee, breakfast sandwiches) to reduce reliance on a single SKU.
Plan for bankruptcy: Had Jack's filed Chapter 11 earlier (when assets still exceeded liabilities), franchisees and suppliers might've been made whole. Lesson: If debt > 2x revenue, consult a bankruptcy attorney before lawsuits hit.
Bonus checklist for F&B operators:
☐ Conduct monthly P&L reviews by location
☐ Survey customers quarterly on product quality
☐ Maintain 3-6 months cash reserves
☐ Audit supplier contracts annually
☐ Test new operational models with 1-2 pilot locations
☐ Hire legal counsel for any fundraising activity
☐ Document franchisee feedback in writing (CYA for disputes)
Future Trends & Unanswered Questions (Forecasting):
Amazon: The Next 24 Months (Q4 2025 - Q4 2027)
Will Anthropic IPO—and at what valuation? If Anthropic goes public at $183B+ (OpenAI-like hype), Amazon's stake could be worth $30-40B.
But if it IPOs at a discount (say, $100B due to profitability concerns), Amazon faces a write-down. Forecast: Anthropic IPO likely in late 2026/early 2027.
Abdulla AL Noman
BzOpa News Pop