Nvidia’s recent decision to scale back certain financing arrangements with emerging Australian artificial intelligence (AI) infrastructure companies has raised concerns about the future availability of innovative funding mechanisms in the growing AI data centre sector. The chipmaker, led by CEO Jensen Huang, had earlier announced a novel business model involving “credit support” deals in which Nvidia provided financial backing to data centre builders Firmus Technologies and Sharon AI in exchange for a share of their revenue. However, the company retired this arrangement last week, less than two months after its introduction, citing competitive pressures and concerns over how much influence Nvidia could exert on customers’ operations.
Firmus Technologies and Sharon AI had been among the first beneficiaries of Nvidia’s experimental financing initiative, which was designed to accelerate adoption of Nvidia’s advanced chips by allowing these companies to access capital at favourable terms through revenue-sharing models. Firmus, co-founded by billionaires Oliver Curtis and Tim Rosenfield, had touted the partnership as a key strategic asset and was reportedly looking at committed offtake deals valued between $25 billion and $30 billion over six years. Nvidia holds a stake in Firmus and had invested fresh capital alongside private equity firm Blackstone in a funding round that valued Firmus at $15.5 billion. Sharon AI, founded in Sydney and serving clients including Canva, operates as a “neocloud” provider by installing and managing computing equipment within existing data centres. Its CEO, James Manning, stated that the company’s situation remains unchanged despite Nvidia’s withdrawal from the financing model.
While the specific credit support for some customers has been paused, Nvidia maintains that its broader financing strategy remains intact and continues to evolve in response to strong demand. The company is actively deploying its financial resources to support the wider AI ecosystem, including major projects such as a planned 10-gigawatt data centre campus in Portsmouth, Ohio, which Nvidia is backing with an estimated $105 billion guarantee. This facility is set to be the largest of its kind globally and will serve major AI developers, including OpenAI, which is expected to purchase up to $350 billion worth of chips over the full build-out.
Nvidia’s CFO, Colette Kress, emphasized that the company’s financial support is essential to sustain rapid growth among frontier AI laboratories, many of which face challenges securing competitive financing to meet soaring demand for computing power. Although some investors have voiced concerns about the potential risks of circular financing—where Nvidia both supplies chips and backs associated debt—others view the approach as an effective means to accelerate the build-out of AI infrastructure and reduce borrowing costs for smaller cloud operators.
The ongoing rollout of such financing initiatives has sparked debate among market participants, with some describing the approach as polarizing due to the complex interplay between Nvidia’s roles as chip supplier, investor, and financier. Nonetheless, Nvidia executives contend that these measures help secure critical inputs and long-term capacity, crucial for sustaining the AI boom. Additionally, the company is addressing infrastructure bottlenecks beyond computing chips, such as power supply, by investing billions across multiple firms to ensure adequate electricity for data centre operations.
Meanwhile, Firmus and Sharon AI remain significant players in the AI infrastructure landscape, benefiting from growing demand for cloud computing and accelerated chips. While Nvidia steps back from some of its experimental credit programs, it continues to support the sector through more traditional investment and financing methods, underscoring the company’s central role in fueling the expanding AI market.