Baillie Gifford US Growth and Scottish Mortgage are now exposed to Hugging Face via Nvidia

From a distance, the announcement looks like just another piece added to the Nvidia empire. Viewed from Baillie Gifford US Growth Trust and Scottish Mortgage Investment Trust, it raises a much more interesting question: Do the managers now own not only the maker of the "shovels and picks" of artificial intelligence, but also one of the places where researchers select, compare, and distribute the models that will power those machines?

Summary in 30 seconds

The fact. On September 2, Nvidia signed a definitive agreement to acquire Hugging Face. The announced price is $12.93 billion; the regulatory filing distinguishes approximately $11.9 billion paid to shareholders and up to $1 billion in stock-based compensation intended to retain employees. Closing is expected in the first half of 2027, subject to regulatory approvals.

The holding company perspective. As of July 31, Nvidia held a 7.0% stake in Baillie Gifford US Growth and a 5.5% stake in Scottish Mortgage. Based on published assets, this represents a combined indicative exposure of approximately £984 million. The acquisition may strengthen its power in the ecosystem, but it also brings developer trust and Hugging Face's neutrality to the heart of the thesis.

The unknown. There is still no proof that the transaction will increase the trusts' revenues, margins, or net asset value. Hugging Face is not yet consolidated, regulatory remedies are unknown, and Nvidia promises to support competing chips.

Why start with holding companies?

As of July 31, 2026, Scottish Mortgage had £16.66 billion in total assets: Nvidia accounted for 5.5%, its third-largest holding after SpaceX and TSMC. Baillie Gifford US Growth had £960.16 million in assets; Nvidia held 7.0% of this, ranking third behind SpaceX and Stripe. Both vehicles thus give their shareholders direct and significant exposure to the acquirer. However, their public documents do not allow us to assert that they held Hugging Face before the announcement: therefore, the transaction should not be presented as a direct exit for the trusts.

The value mechanism operates through Nvidia: quality of allocated capital, increased distribution power, ability to stimulate computing demand-and only then, in the long term, operating profit and share price. A hypothetical 10% change in Nvidia's share price would represent, all other things being equal, approximately 0.70% of Baillie Gifford US Growth's assets and 0.55% of Scottish Mortgage's. This sensitivity analysis is neither a forecast nor a measure of instantaneous net asset value.

What Nvidia is really buying: a "shelf" that has become infrastructure

Hugging Face is often described as the GitHub of artificial intelligence. The image is imperfect, but fruitful. Its platform brings together more than three million models, 500,000 datasets, and one million applications, used by more than 200,000 companies, according to Nvidia. This catalog is not simply a repository: it influences model discovery, evaluation, documentation, and deployment.

Nvidia already dominates a crucial physical layer-accelerators and the software that makes them usable. With Hugging Face, it is moving closer to a distribution layer where models, data, developers, and compute providers converge. The strategic benefit is circular: more easily deployable models can boost inference; more available compute makes the platform more useful; Observing usage patterns can finally help Nvidia guide its libraries, services, and architectures.

This loop is a Inference, not a quantified commitment from Nvidia. The company has not published any synergies, expected contribution to revenue, or integration timeline. However, it has indicated that it is already the leading contributor to Hugging Face with some 500 models and 250 datasets. The transaction thus formalizes a close relationship that existed before the check.

The paradox: Nvidia shouldn't "own" Hugging Face too much

The value of Hugging Face lies precisely in the fact that a developer can find several models, several frameworks, several clouds, and several hardware architectures. Nvidia has committed to maintaining this choice: the use of its chips will not be mandatory, and other silicon suppliers will remain supported.

This is the crux of the matter for the investor in the two trusts. If Nvidia transforms the platform into a sales funnel for its own products, it could damage its traffic, diversity, and the trust that justified the acquisition. If it truly maintains neutrality, it will have to accept that some of the value will benefit competitors. The best asset, therefore, is the one its new owner knows not to squeeze too hard.

This paradox also explains the regulatory risk. The document filed with the SEC acknowledges that governments could restrict certain open models, particularly those originating in China, and that such measures could materially affect the platform's business. The review will likely also focus on non-discriminatory access for competing providers. Here again, this is an analytical interpretation; the authorities have not yet published specific conditions.

A significant purchase, but not transformative in its size

The $12.93 billion price tag makes headlines. For a company the size of Nvidia, however, the real question isn't "Can it pay?" but "Can it earn a return exceeding its cost of capital without breaking the asset?" The breakdown is helpful: approximately $11.9 billion in acquisition costs, plus up to $1 billion in stock options to retain employees. This last component serves as a reminder that the product is also a community and a team, not just a database.

With the deal not expected to close until the first half of 2027, it would be premature to call the acquisition "completed." Until then, the two trusts hold Nvidia stock based on a promise of integration, not on Hugging Face's consolidated results. The process involves separating three levels: the signed agreement, the strategic potential, and then the financial proof. Currently, we only have the first and plausible arguments for the second.

The value loop depends on four conditions

Each arrow represents a condition. Developers must stay. The models must be used in production. Nvidia must monetize without shutting down. And the price paid must remain below the present value of those earnings. It is precisely these conditions-and not the raw size of the catalog-that will need to be tested.

Two trusts, two shades of interpretation

For Baillie Gifford US Growth, Nvidia is a significant public holding within a portfolio that also includes nearly 30% private assets. The transaction can illustrate how a publicly traded champion acquires community infrastructure to extend its growth trajectory. But it also increases the portfolio's thematic overlap: chips, models, and platforms all depend on the same AI investment cycle.

Scottish Mortgage is much larger and owns, alongside Nvidia, TSMC, ASML, and Anthropic. The acquisition, therefore, doesn't necessarily diversify its engine; it densifies its AI value chain. This is potentially powerful if end-use becomes widespread, but more fragile if computing expenditure slows simultaneously across multiple levels. The actual economic concentration may be greater than a line-by-line analysis of the portfolio suggests.

What to watch now

The real find, then, isn't that Nvidia is buying a popular site. It's that a computing power provider is trying to acquire the common language of those who consume it. For shareholders of Baillie Gifford US Growth or Scottish Mortgage, the event is noteworthy because it can extend the growth trajectory of one of their largest positions. It also warrants vigilance because the main asset being acquired-the trust of an open ecosystem-doesn't appear on any balance sheet and can disappear faster than a factory.