Ex-a16z GP Midha's Amp Commits $1.3B Bundled With Compute

Amp, led by former Andreessen Horowitz general partner Anjney Midha, has committed about $1.3 billion to AI ventures alongside pooled GPU access.

hedgeco.net reports that Anjney Midha, a former Andreessen Horowitz general partner and early Anthropic investor, has advanced Amp with about $1.3 billion committed for a venture model that bundles investment capital with pooled access to scarce AI compute. According to hedgeco.net, the September 2026 coverage from Gokhshtein and Startup Fortune, plus earlier reporting in outlets including AI Business, frames the structure as pairing investment capacity with an infrastructure layer intended to secure and allocate GPU access. hedgeco.net is careful with the label: sources describe more than $1.3 billion committed, rather than a universally labeled traditional final close, and no full LP roster beyond what those pieces disclose is claimed.

The consensus-breaking move here is Amp's thesis, as framed in the hedgeco.net summary of the coverage: chip and cluster scarcity, not the amount of committed capital, is the binding constraint for frontier AI startups and researchers. Traditional venture vehicles compete on check size; Amp instead positions itself as both a capital partner and a compute allocator, effectively turning GPU allocation into part of the product a founder buys when they take the money.

hedgeco.net grounds the figures in named September 2026 coverage from Gokhshtein and Startup Fortune, with earlier reporting in outlets including AI Business, and it instructs careful attribution: 'about / more than $1.3 billion committed,' the compute-plus-capital bundling, and Midha's a16z and Anthropic background are all sourced claims. The outlet also flags that separate reporting has noted Amp-linked capital into Anthropic financings, but says those should be treated as context only unless a specific ticket is required. Hard details readers might expect — a final-close date, a full LP roster, or a fund vintage label — are deliberately not claimed, because the underlying wires do not use them.

If Amp can reliably secure and allocate GPU access, the model could pressure conventional funds to compete on infrastructure rather than term sheets, and give frontier AI founders a reason to pick a smaller allocator with cluster capacity over a larger check without compute. Watch for a traditional close announcement or named LP disclosures as the next confirmation point — and, per Stock Titan's TriplePoint coverage, for other platforms adding investment-leadership roles to chase the same AI infrastructure demand.