AAPL: The Business Passed Every Test. The Price Failed Four Ways.

On September 30, 2026, the BagsCapital Investment Committee returned Watch, Avoid, Watch, Watch on Apple. Zero Buys. The published selection rule needs three of four Buy verdicts before a name enters any basket. Apple got none.

All four seats praised the business. All four declined the price. The interesting part is that the refusals are not the same refusal.

The session landed on a morning with two catalysts stacked on one name. Bloomberg reported that Apple's long-delayed smart home launch arrives October 13, the first new product category in years and the first big test for a CEO who took over September 1. A day earlier, Apple shares fell more than 2.5% after a major US bank's research note argued that AI agents can take discovery, referral, and transaction initiation away from the ecosystem even if Apple keeps every handset sale.

The fact sheet every investor read

The missing rows are a finding, not a gap. We did not fill them with assumptions. That matters here, because the one threat the whole market is arguing about, who owns the transaction, lives exactly in the rows the filings do not give us.

How Apple makes money

Added September 30, 2026. The four sections that follow were not in the first version of this article. Later on September 30, the committee's analyst, which does not vote, and all four investors were given the same fact sheet again and asked the questions this format now asks: how the business works, what can break it, which environments decide it, and what has to be true. The verdicts in this article are the ones recorded at the session. Asked again, Investors A, B and C repeated theirs. Investor D returned Hold rather than Watch.

Apple sells devices, the iPhone above all, plus iPad, HomePod and Apple TV, and runs the ecosystem those devices sit in: the App Store and the channel through which users discover products, follow referrals and start transactions. Revenue was $466.82B over the twelve months to June 27, 2026, with net income of $128.93B, a 27.6% net margin.

The fact sheet cannot split that revenue between hardware and services. The analyst flagged why that matters: the agent argument in the bank note "is aimed at the services and ecosystem layer, not at the handset." Segment revenue, the Services share, unit sales and revenue per user are all not available.

The business funds itself. Last fiscal year's capital spending was about $12.71B, roughly 3% of revenue, and free cash flow over the trailing twelve months was $136.68B. Apple holds $146.52B in cash and short-term investments against $82.35B of debt, which leaves $64.17B of net cash. The analyst noted one sign of operating leverage: trailing revenue rose $50.66B over the last fiscal year while net income rose $16.92B, an incremental margin of about 33%, above the 27.6% average.

The analyst's summary of the moat: "The moat the sheet describes is control of the device and the installed base. The contested ground is who owns the user's intent once the device is in their hand."

The risks to Apple

The analyst ranked the risks that could permanently impair the business, not the ones that only move the stock.

The analyst's most underappreciated risk is the second one. The Google reliance "appears in a single closing sentence" of the news summary, yet it sits under both headline stories: it powers the new Siri at the center of the home launch, and it is the capability Apple needs to answer the agent threat. "If Apple's answer to one outside platform taking over the user's intent runs on another outside platform, then the company controls less of its own defense than the installed-base argument suggests." The sheet gives no terms or duration for the arrangement, so the risk cannot be sized.

The next two to five years

We added this part later on September 30, 2026. These are the risks that would not end Apple's franchise but could weigh on earnings, or on what investors pay for them, for years.

What the fact sheet did not cover

The first version of this article did not mention Taiwan, because the fact sheet did not. That was a gap in the sheet, and it is the biggest one. What follows comes from the analyst's general knowledge rather than the sheet, so there are no figures. Apple's processors are made by TSMC, mainly in Taiwan, and at the leading edge there is no practical substitute, which makes cross-strait risk a single point of failure. Final assembly is still concentrated in China, with a growing share in India and Vietnam, so US and China tensions, tariffs and export controls all reach the product. Google pays Apple to be the default search engine, and remedies in the US antitrust case against Google could restrict that. App Store rules are being rewritten by the EU's Digital Markets Act, Japan's smartphone competition law and the Epic litigation, while the US Justice Department pursues its own antitrust suit against Apple. The next Apple fact sheet should carry the share of chips made outside Taiwan, assembly by country, the size of the search payment, and commission rates by region.

Investor A, quality and long-term value

He works on a 5 to 20 year horizon. His opening judgment split the business from the stock: "I would own this business. At today's price I would not buy it."

He praised the economics. A 27.6% net margin on hardware, trailing free cash flow above trailing net income, $64.17B of net cash. His objection was the spread: a 2.84% cash yield against a 5.24% Treasury yield. An owner accepts "about 240 basis points less cash yield" than a bond, then needs growth to close that gap for decades.

The agent question, he wrote, "goes to the moat itself, not to the next quarter." Without Services revenue on the sheet he cannot measure the exposure at all.

He also declined to take the cash flow jump at face value. Free cash flow rose from $98.77B to $136.68B while revenue grew 12.2%. His caution: "part of the trailing jump could be working-capital timing."

His entry discipline is 20x to 22x normalized free cash flow, roughly $187 to $206 per share. Valuation arithmetic, not a price prediction. His own strongest counter cuts against his caution: "whoever controls the gate can often set the terms for those who pass through it."

Verdict: Watch. Conviction 7 of 10.

Investor B, deep value and margin of safety

He works on a 1 to 5 year horizon, and he returned the session's only Avoid: "This may be a good business, but it is not a good investment at this price."

His arithmetic is the bluntest thing in the transcript. Applying 20x to 25x to last full-year earnings of $112.01B gives $2.24T to $2.80T of equity value. The market cap is $4.81T. In his words, a "72% to 115% premium, rather than a discount."

The October 13 products do not move his number. He assigned them no value "until their economics are demonstrated."

His purchase threshold is at or below roughly $1.57T of equity value, a 30% discount to his own conservative case. A conditional purchase threshold, not a price forecast. His own counter is that "last full-year earnings may materially understate sustainable future earning power."

Verdict: Avoid. Conviction 8 of 10.

Investor C, global macro and reflexivity

He works across days to years, and he objected to what the market is doing rather than to what Apple is doing: "The market appears to be pricing Apple as safety. It is not safety at this price, it is duration."

The roughly 240 basis points of negative carry is, in his phrase, "uncompensated duration risk dressed as quality." His regime label: "a tightening-complacency regime." The 10-year rose 10.1% in 30 days while the S&P 500 sat flat, which tells him equities have not repriced the move in yields at all.

On the agent note he made the sharpest structural point of the session. Annual software cadence against agents that improve daily is "a tempo mismatch, not a one-day headline," and the market absorbed it "as noise rather than repricing it."

His crowding tell is blunt. The month's most prominent bear argument still ended in a Buy and a $370 target. "When the bears are long, the distribution is one-sided."

His entry conditions are conditions, not levels: the 10-year back below its 30-day starting area near 4.76 and stabilizing, the cash yield recovered to the mid-3s, an October 13 that shows Siri holding its own, and services evidence of resilience. His own counter: "The fundamentals are doing exactly what bulls say they are doing," and "crowded can stay crowded for years."

Verdict: Watch. Conviction 6 of 10.

Investor D, growth and stock picking

He works on a 2 to 10 year horizon, and his variant perception is the reason this session is worth reading. The consensus, he wrote, "assumes Apple will automatically win the edge AI transition because it owns the consumer hardware." His read: "user interaction patterns are decoupling from the underlying operating system."

Keep every device sale, lose where the transaction starts. He called that "a structural threat to Apple's long-term margin expansion."

He matched C on tempo from the growth side. Apple ships major software once a year while agents improve daily, and the new Siri lacks persistent background tasks, broad third-party action support, and a dedicated agent payment rail.

His valuation point is the growth seat's version of the same refusal. At 37.3x trailing earnings against 10% to 12% growth, the PEG exceeds 3.0x, with no cushion for missteps. His entry discipline is 20x to 24x earnings, a PEG nearer 1.8x to 2.0x. His own counter is the mirror of his thesis: if privacy trust wins, Apple becomes "the only trusted AI orchestrator," growth re-accelerates into the high teens, and the premium is justified.

Verdict: Watch. Conviction 7 of 10.

The environments that decide Apple

Each investor named the operating conditions that matter most to this business, and the one that would change their verdict.

Three of the four investors named the same critical environment, the fight over who owns discovery and transactions. Investor A: agents visibly capturing that layer "would move my verdict from Watch to Avoid at any price near today's, because it attacks durability and not just growth." Investor B: proof that the transition produces measurable new earnings could move Avoid to Watch. Investor D: Siri becoming the primary orchestrator of intent would mean Buy, agents bypassing the App Store would mean Avoid.

Investor C alone put rates first, and said so in numbers: the 10-year holding above roughly 5.5% moves Watch to Avoid or Sell, a turn toward the mid-4s moves it toward Buy.

What has to be true for Apple

The bull case, link by link, from the economy down to a single user.

The weakest link split three ways, and the split is the finding. Investors A and D chose the industry link, the agents. Investor A paired it with the one number that would settle it: "Link 10 is the number that would confirm or refute the threat, and it is not available." Investor C chose rates: "The entire chain is priced off a multiple that only survives if the discount rate cooperates." Investor B chose unit economics: "consolidated cash generation does not establish the profitability of the next growth initiative."

Investor A drew the line between them: "Link 1 is contradicted too, but it is a price problem, not a business problem. Price problems can fix themselves, while a lost moat usually does not come back."

Where they split on Apple

The verdicts look uniform. The reasoning is not.

A refused on margin of safety against a threat he cannot measure. B refused on evidence, the premium itself. C refused on regime, arguing the discount rate matters more than the cash flow. D refused on growth, arguing the multiple already prices flawless execution. Three of the four anchored on the same spread, 2.84% against 5.24%, and refused it three different ways.

That separation is what makes the verdicts testable. Better services disclosure would let A size the agent threat. Demonstrated earning power would meet B's objection. A 10-year that stabilizes would satisfy one of C's conditions. Evidence that Apple can own the agent relationship would answer D.

The unplanned convergence sits underneath all of it. A's entry arithmetic lands at $187 to $206. B's $1.57T threshold is roughly $107 per share. Both sit far below the $329.40 tape, from opposite philosophies. We saw the same pattern in our CarMax review the day before, and a version of the same refusal in our Meta review. These are conditional entry calculations. They describe each seat's discipline, not where shares will trade.

What we take from the Apple review

Apple's business quality survived the meeting. Its price failed all four tests, for four different reasons.

The most useful output is the missing data. Without services disclosure, the committee could name the agent threat but could not size it. Every seat treated that as a finding rather than a gap to estimate into, which is the entire point of the method. A number nobody can decompose is not evidence.

The October 13 launch gives everyone a dated event to watch. It does not settle who owns the transaction when a user talks to an agent instead of an app. That question is now on the record, with the conditions each seat would need to change a vote.

Under our three-of-four rule, Apple enters no basket.

As of September 30, 2026, BagsCapital manages $0 in assets under management and holds no position in Apple.