META: Four Verdicts, Zero Buys, One $2 Trillion Question

On September 25, 2026, the BagsCapital Investment Committee returned four verdicts on one Meta Platforms fact sheet: Watch, Avoid, Watch, Watch. Zero Buy verdicts. The published selection rule requires three of four Buy verdicts before a name can enter any basket. Meta did not qualify.

The session came the day after Bloomberg reported Meta’s best month since 2013 and a market value approaching $2 trillion. All four seats praised the business. None accepted the investment case at the supplied price.

We think the reasons matter more than the tally. One seat wanted better cash conversion. Another could not establish intrinsic value from the evidence. A third distrusted the timing of the narrative. Even the growth investor set an entry range well below the tape.

The Meta Platforms fact sheet every investor read

The filing record shows earnings rising while free cash flow fell. That comparison uses fiscal 2025 and the trailing twelve months to June 30, 2026, which are overlapping periods.

Investor A, quality and long-term value

A works on a horizon of 5 to 20 years. His opening judgment: “This is one of the finest businesses I have looked at.” Then he separated admiration from ownership economics. “The question is not quality. It is what an owner actually keeps, and what they pay for it today.”

We read his Watch as a demand for demonstrated cash conversion and a price that leaves a margin of safety. Only 60.2% of trailing net income became free cash flow. His stated buying discipline was an earnings yield of 6% to 7%. He described that as roughly 15x to 17x trailing net income and a market value near $1.0 to $1.15 trillion. Those are his conditions for interest. They are not forecasts.

His uncertainty cuts directly into his method: “I may be treating growth capex as though it were a permanent cost.” If today’s spending creates future owner cash, his present measure may be too restrictive. He still concluded: “I want to own this business. I do not want to buy it at this price.”

Verdict: Watch. Conviction 7 of 10.

Investor B, deep value and margin of safety

B’s horizon is 1 to 5 years. His refusal began with what the sheet could establish: “My primary reason is the absence of a verifiable margin of safety at the supplied price.” Normalized earnings were unavailable. So were tangible book and liquidation value. Within his method, he could not compute intrinsic value from the supplied record.

This is a different objection from A’s. A could state terms under which ownership interested him. B could not verify the valuation needed to justify a purchase. “Accounting earnings alone could materially overstate the cash available to owners after investment.” Missing evidence did not mean the business lacked economic value. It meant his purchase test could not be completed.

His rule was absolute: “No verifiable margin of safety means no purchase.” So was the boundary around that conclusion: “This is a rejection of the investment case at this price, not a prediction of a decline.”

Verdict: Avoid. Conviction 8 of 10.

Investor C, global macro and reflexivity

C works across days to years. He read the sequence as a feedback loop: “price rose first, then the Muse launch on September 8 and the Connect hardware showcase confirmed the narrative”. His interpretation of the subsequent coverage was equally direct: “milestone coronation coverage is itself a late-cycle tell.” These are judgments about market behavior. They are not filing facts.

The rate backdrop supplied his harder constraint. The US 10-year Treasury yield stood at 5.18%. It had risen 11.6% over 30 days. Bloomberg also reported a sell-side warning about hyperscaler debt supply. C connected the two: “The buildout is starting to compete for the very capital that sets these companies' discount rates.” We see the same financing question running through our earlier discussion of AI buildout exit conditions.

His Watch was a judgment about entry timing and verification. The product catalysts had arrived without a new earnings report or guidance change in the preceding 14 days. His own objection to waiting deserves equal space: “crowded can stay crowded for quarters while I wait for a washout that arrives late and small.”

Verdict: Watch. Conviction 6 of 10.

Investor D, growth and fundamental stock picking

D’s horizon is 2 to 10 years. He was the most bullish seat on the growth case. Yet his assessment was that “the market has already priced in near-term perfection for unmonetized artificial intelligence products”. His cash concern was specific: “artificial intelligence datacenter investments are absorbing capital faster than cash from operations grows.”

His stated entry range was $580 to $640. That sits 17.7% to 25.4% below the supplied $777.59 price. We treat that range as his buying discipline. It is not a forecast. He still returned Watch.

His failure case was also specific. He would be wrong if Muse and the Muse Charm triggered a massive unexpected cycle of high-margin hardware sales and subscription revenue. His method leaves room for that outcome. The supplied evidence did not establish it.

Verdict: Watch. Conviction 7 of 10.

Where they split on Meta Platforms

Nobody disputed the business. All four praised it. The disagreement concerned what price and evidence could justify a purchase. That distinction matters because three Watch verdicts can look like one shared conclusion. They were not.

A wanted proof that more earnings would become more owner cash. He also wanted a margin of safety in the price. B refused an earlier step in the process. The sheet could not support an intrinsic value calculation under his framework. C thought the narrative had already received its catalysts while the rate backdrop worked against the entry. Cash conversion, verifiability and timing are three separate objections.

The sharpest shared fact was the divergence between net income and free cash flow. Income rose 12.6% across the supplied comparison. Free cash flow fell 11.1%. Against that stood a September rally reported at about 36% by GuruFocus, the weaker source. Our reading is that the rally priced the cash line as though it had followed the earnings line. That is an interpretation of the mismatch. It is not proof of what every buyer believed.

D makes the distinction between business and price hardest to evade. Even the growth seat placed his entry discipline roughly 18% to 25% below the tape. Enthusiasm for the products did not produce a Buy.

The missing fields also matter. The official P/E was unavailable. A roughly 29x earnings multiple comes from dividing Bloomberg’s headline market value of about $2 trillion by trailing net income. That is arithmetic on a headline, not a filing figure. Short-term and current debt were also unavailable. The stated $6.60B net cash figure may therefore overstate the cushion.

We would not let weaker inputs close those gaps. The Amazon block remains an unconfirmed rumor. The board discussion records sentiment. Neither supplies the missing valuation evidence or demonstrates product monetization.

What we take from it

A reported 36% month repriced the narrative while the cash line moved the other way. All four seats confronted that divergence because the same fact sheet put the same numbers in front of four different philosophies. The method made the disagreement visible.

It did not prove the committee right. A may be penalizing spending that later earns its keep. B’s evidence threshold may exclude opportunities before their value becomes verifiable. C may wait through a rally that persists. D may underestimate a new revenue cycle. A useful refusal names what could overturn it.

We saw a similar outcome in the earlier Nvidia committee session. Four seats read a great business at a demanding price. Not one said Buy. The repeated tally matters less than whether the reasons remain distinct and open to challenge.

That is the committee’s job. Make the refusals visible. Keep three different refusals from collapsing into one generic valuation concern. The three-of-four rule then applies to the actual verdicts. On this sheet, it had zero Buy votes to count.

As of September 25, 2026, BagsCapital manages $0 and holds no position in Meta Platforms.