GAVIN BAKER: Funded commitments such as the $35 billion Apollo and Broadcom deal are taken seriously, and critics conflate them with unfunded pledges

The Gist

Gavin Baker says Ed Zitron's examples are deals that already have money behind them, and of course suppliers take those seriously. His point was about the much larger pile of promises with no money behind them. This is a steelman: each side is reconstructed in its strongest fair form, not endorsed, and no winner is declared.

Conclusion

Pointing to funded deals does not rebut the claim that unfunded commitments are discounted, because funding, not announcement size or legal form alone, is what the supply chain treats as decisive.

Premises

1. A commitment with external financing in place, such as the $35 billion Apollo and Broadcom arrangement, is funded by definition.
2. The supply chain takes funded commitments seriously and builds against them.
3. Smaller funded deals (about $35 billion and $18 billion) are categorically different from the roughly $1 trillion of unfunded commitments, even if some unfunded ones are non-cancellable.
4. Citing funded deals to rebut a claim about unfunded commitments mixes the two categories.

Assumptions

• Funded deals are a real but much smaller share of the headline total.
• Broadcom backstops part of the Apollo financing (up to $29 billion on the first tranche), so "funded" here includes supplier credit support.
• The $18 billion figure Zitron cited is a bank project loan for Oracle's New Mexico site, not bonds; Oracle separately sold $18 billion of corporate bonds in September 2025.

Analysis

Argument type: Deductive.

Potential Fallacies

• Unsupported generalization about collective behavior (P2 and the conclusion) — 'The supply chain' is treated as one actor following one rule. Foundries, memory makers, power providers, and data-center builders likely weigh counterparty credit, contract terms, and strategic relationships differently. No sample of supplier behavior is offered, so the claim that funding is decisive rests on the speaker's authority and general plausibility.
• Non sequitur (sufficient vs. decisive condition) (P2 to the conclusion) — P2 says funded commitments are taken seriously, which makes funding a sufficient signal. The conclusion needs funding to be the decisive or necessary signal, so that unfunded commitments are discounted and legal form matters less. The leap from 'funded implies serious' to 'only funded is serious' needs a bridging premise that is not supplied.
• Stipulative definition / near question-begging (P1, P4, and the conclusion) — Calling a deal 'funded by definition', with supplier credit support included under A2, settles the classification by stipulation. Whether the classification predicts supplier behavior or commitment durability is an empirical question. P4 also restates much of the conclusion's first half, so the argument is partly true by construction and says little about the real dispute.
• False dichotomy (funded vs. unfunded) (P3 and P4) — Real financing runs along a spectrum: committed and undrawn tranches, conditions precedent, partial backstops, and non-cancellable contracts from strong-credit counterparties. P3's own concession about non-cancellable unfunded commitments blurs the 'categorical' line that P4 relies on.
• Possible straw man / relevance gambit (P4, title, and A3) — The argument charges critics with conflation without quoting their claim. If the critic was arguing that headline totals are overstated, or that financing is fragile or circular, then citing funded deals is not a category error. Correcting one mislabeled $18 billion instrument also does not show that the critic's broader thesis is wrong.
• Minimizing by comparison (P3 and A1) — Setting $35 billion and $18 billion beside $1 trillion makes the funded deals look negligible, but the ratio does not show how suppliers actually discount anything. The comparison also treats the $1 trillion as one homogeneous 'unfunded' pool, though it mixes non-binding MOUs, non-cancellable contracts, and commitments backed by hyperscaler cash flow.

Counterarguments

These challenges are not ranked.

• Premise 1 (High impact) — A Broadcom backstop means the supplier is credit-enhancing its own customer's purchase. That is vendor financing, and it resembles the circular financing skeptics already allege. The 'funded' exemplar may therefore be the weakest evidence of independent demand, and could even support the critic.
• Premise 2 (High impact) — Suppliers routinely build against non-cancellable take-or-pay contracts from strong-credit counterparties, such as hyperscalers, without external financing. If funding were truly decisive, those contracts would be treated as worthless, which is implausible. Counterparty credit and enforceability may matter more than the funding label.
• Premise 3 (High impact) — The stated figures imply that only about 5% of the headline total is funded. Even granting the category distinction, that supports the critic's thesis that headline totals are overstated. The 'unfunded' bucket also hides large differences in reliability, including commitments that will likely be financed later in capital markets.
• Premise 4 (Medium impact) — The critic may have cited funded deals to show the scale of the buildout, or to point out financing fragility, not to rebut the discounting claim. In that case the conflation charge is a straw man, and the funded examples are legitimate evidence on the broader question.
• Conclusion (Medium impact) — The narrow point wins on relevance but leaves the contested question untouched: how much of the unfunded pool will actually materialize, and whether suppliers discount it correctly. Funding status is also dynamic. Unfunded commitments can become funded within supplier lead times, and funded anchor deals can raise the odds that adjacent commitments get financed.

Suggested Improvements

• Evidence for supplier behavior — Cite observable supplier actions, such as capacity reservations, prepayments, capex guidance, and backlog definitions, showing that suppliers build against funded commitments and discount unfunded ones, and compare them with non-cancellable unfunded contracts. P2 carries the stronger conclusion but is currently asserted, not shown.
• Decompose the ~$1 trillion — Break the unfunded total into non-binding MOUs, non-cancellable contracts, and commitments backed by investment-grade balance sheets or operating cash flow. This would show how much of the pool is truly discountable and address the concession in P3 directly.
• Treatment of the backstop — Explain why a supplier-backstopped financing counts as independent funding, and say who bears first-loss risk. Alternatively, distinguish third-party funding from vendor-supported funding. This removes the strongest circularity objection and sharpens the 'funded' category.
• Quote the critic's claim — Reproduce the critic's actual claim and show that the funded deals were offered as a rebuttal to a claim about unfunded commitments. The conflation charge depends on what was actually argued, and quoting the claim guards against a straw-man reading.
• Narrow or bridge the conclusion — Either limit the conclusion to the relevance point, or add an explicit premise that suppliers discount in proportion to lack of funding regardless of legal form, and then defend it. The strong 'funding is decisive' claim currently outruns the premises.

Scenario Tests

• A critic cites the $35 billion Apollo/Broadcom deal as evidence that the overall AI commitment pool is fully credible. (Supports) — In this case the category point applies. One funded deal does not show that the roughly $1 trillion of unfunded commitments carries equivalent credibility, and the critic would be making the conflation Baker describes.
• A critic cites the funded deals to show that only a small fraction of headline commitments is actually financed. (Challenges) — Here the funded examples are relevant, because the small funded share supports the critic's aggregate point. The conflation charge misses, and the argument answers a different question than the one asked.
• A foundry or memory maker expands capacity against a non-cancellable take-or-pay contract from a strong-credit hyperscaler that has no external financing. (Challenges) — Suppliers would be building against an unfunded commitment, so legal form and counterparty credit would be doing the work, and 'funding is decisive' would fail for part of the pool.
• The Broadcom backstop is drawn on, or the Apollo financing proves contingent on conditions that fail during a credit tightening. (Challenges) — The 'funded' exemplar would turn out to be conditional and dependent on the supplier's balance sheet, which blurs the funded/unfunded line and weakens funding as a safety signal.
• Supplier order books and capex guidance track funded commitments closely, while unfunded announcements produce no prepayments or capacity reservations. (Supports) — This would give P2 the behavioral support it lacks and make the discounting claim empirically credible, though it would still need controls for counterparty credit and contract terms.

Coherence & Relevance

The argument is coherent as a relevance rebuttal. Its first half (funded examples do not directly answer a claim about unfunded commitments) follows from P3 and P4, and the factual context in A1 to A3 is consistent with it. Its second half (funding is what the supply chain treats as decisive) is not established, because the only support is the unevidenced P2, and the argument's own concessions about non-cancellable commitments and supplier backstops erode the clean binary it depends on. The result is a sound narrow point wrapped in a broader empirical claim that remains unproven.

• A commitment with external financing in place, such as the $35 billion Apollo and Broadcom arrangement, is funded by definition. (Moderate) — It establishes that the exemplar deal belongs to the funded category but does not link funded status to supplier behavior. Including the supplier backstop under 'funded' leaves open whether the supply chain treats vendor-supported financing like independent financing.
• The supply chain takes funded commitments seriously and builds against them. (Strong) — It is the key link to the conclusion, but it states only that funded commitments are taken seriously. It does not say that unfunded ones are discounted or that funding outweighs legal form, so a bridge is missing.
• Smaller funded deals (about $35 billion and $18 billion) are categorically different from the roughly $1 trillion of unfunded commitments, even if some unfunded ones are non-cancellable. (Strong) — It supports the relevance point well, but its non-cancellable concession is in tension with the dismissal of legal form. No limiting principle explains why binding but unfunded commitments are still discounted.
• Citing funded deals to rebut a claim about unfunded commitments mixes the two categories. (Strong) — It nearly restates the first half of the conclusion and depends on an unshown characterization of the critic's actual claim.

View this argument on LogicFirst.ai