- Hyperscale players such as Meta Platforms, Alphabet Inc. and Microsoft Corporation expect to generate only US$1.4 trn from cash flows while needing roughly US$2.9 trn in capex by 2028.
- The funding strategy includes vendor‑financing, asset‑backed leases and private‑credit vehicles rather than traditional bank loans and standard corporate bonds.
What happened: Innovative financing bridges $1.5T AI gap
According to a report by Morgan Stanley strategists, the major tech companies anticipated to build the bulk of global AI infrastructure expect capital expenditures of about US$2.9 trn by 2028, but believe only around US$1.4 trn will be covered by their own internal cash flows.
To fund the remaining roughly US$1.5 trn gap, firms are deploying novel financing structures. For instance, Meta’s upcoming “Hyperion” data‑centre project in Louisiana is being built by a joint venture in which an asset‑manager controls 80 % and Meta holds 20 %, while Meta leases the facility when completed. The debt and lease structure allows Meta to keep most liabilities off its balance sheet.
In parallel, AI‑hardware supplier NVIDIA Corporation is backing cloud infrastructure provider CoreWeave Inc. via equity and special‑purpose purchase commitments — effectively using vendor financing to ensure demand for its chips while enabling the build‑out of AI compute capacity.
Also read: Vodafone and Three to overhaul infrastructure for smarter growth
Also read: Vodafone launches $545M buyback after growth
Why it’s important
The shift in financing models signals that the AI build‑out is no longer just a technology story but a major financial engineering challenge. Firms are moving away from pure cash‑funded investment and instead embracing off‑balance‑sheet structures, private credit and vendor‑financed ecosystems.
This matters because those structures carry risks: when leverage is high, returns must justify the investments. Analysts point to parallels with past infrastructure booms — and caution that if AI model revenues don’t scale, firms could face overcapacity or write‑downs.
Moreover, Wall Street and asset‑managers now have a growing appetite for backing AI infrastructure through opaque special‑purpose vehicles and non‑bank credit facilities, which expands the types of entities responsible for underwriting AI capital risk.
For investors, the shift raises questions about where future profits will come from: are companies building enough new revenue streams to justify such spending? Some analysts warn that while core tech players have strong cash flows, the private‑credit‑backed firms and vendors may be more exposed.
Overall, the evolving playbook for funding AI infrastructure highlights that the bearers of both opportunity and risk extend far beyond traditional technology budgets and now include the financial services ecosystem itself.
