VC
Value Add VC
โšกHomePulseโšกHelpful Apps๐Ÿ“Blog
โ† Value Add PulseAI-3.39% on July 13

Nvidia's $1 Trillion Wipeout Since Its May Peak

Nvidia fell another 3.39% on July 13, extending a slide that has erased roughly $1 trillion in market value since its May 14 record high, as hyperscalers accelerate a shift toward custom ASIC chips.

-3.39%
July 13 move
~$1T
Value lost since peak
May 14, 2026
Peak date
~18x
Forward P/E now
~21x
S&P 500 forward P/E
TC
Trace Cohen
Early-stage VC & angel ยท Founder, New York Venture Partners
July 13, 2026
2 min read
ShareXLinkedInEmail
THE RUNDOWN
1

NVDA fell 3.39% on July 13, extending a decline that has now erased roughly $1 trillion in market value since the stock's record high on May 14, 2026

2

Nvidia now trades around 18x forward earnings, below the S&P 500's roughly 21x multiple -- the cheapest the stock has traded relative to the index since early 2019

3

Reports that major hyperscale cloud customers are accelerating a shift toward custom ASIC chips to reduce reliance on third-party hardware are a key driver, alongside hawkish Fed signals raising cost-of-capital concerns for high-growth tech broadly

4

Additional pressure points include gross-margin compression from higher memory costs, increasingly crowded institutional ownership of the stock, and criticism that Nvidia is deploying cash toward vendor financing of its own customers rather than buybacks or dividends

TC
The VC Read ยท Trace's TakeTrace Cohen

An 18x forward multiple on the company that still sells the vast majority of AI training compute is either the cheapest AI stock in the market or the market correctly pricing in that its biggest customers are about to become its biggest competitors. I don't think Nvidia's moat evaporates this fast, but the fact that its own customers are the ones building the threat is a different risk profile than any prior chip-cycle downturn, and it deserves to be priced differently than a standard cyclical dip.

Nvidia shares fell 3.39% on July 13, extending a slide that has now erased roughly $1 trillion in market value since the stock's record high on May 14 -- one of the largest sustained drawdowns for any single company in market history, made more notable by the fact that it's happening at a company whose underlying revenue growth hasn't visibly cracked. NVDA now trades around 18 times forward earnings, below the S&P 500's roughly 21x multiple and the cheapest the stock has traded relative to the broader index since early 2019.

The drivers are a mix of macro and company-specific. Hawkish signals from the Federal Reserve have raised cost-of-capital concerns across high-growth technology broadly, hitting high-multiple growth names harder than value stocks almost by definition. But the more Nvidia-specific driver is reporting that major hyperscale cloud customers -- the same companies whose capex has been the core bull case for the stock -- are accelerating a shift toward custom ASIC chips designed to reduce their reliance on third-party hardware like Nvidia's GPUs.

That ASIC shift isn't hypothetical: Meta's own MTIA-based "Iris" chip, designed with Broadcom and manufactured by TSMC, is set to enter production in September and is explicitly intended to help scale Meta's own compute capacity, reducing (at the margin) how much of that scaling has to run through Nvidia purchases. Google, Amazon and Microsoft have all pursued similar in-house silicon programs for years with varying degrees of success; what's changed is that investors now appear to be pricing in meaningfully faster hyperscaler ASIC adoption than they were even a few months ago.

Additional pressure comes from gross-margin concerns tied to higher memory costs -- the same high-bandwidth memory supply crunch that helped drive SK Hynix's record US IPO earlier this year -- alongside increasingly crowded institutional ownership of NVDA and specific criticism that Nvidia is using its substantial cash generation for vendor financing arrangements with its own customers rather than returning capital through buybacks or dividends, a structure some investors see as propping up demand rather than reflecting organic order strength.

For infrastructure investors, the drawdown is a real-time test of whether Nvidia's moat -- its CUDA software ecosystem, training-cluster performance lead and manufacturing scale advantage -- is durable enough to absorb genuine hyperscaler ASIC competition without a lasting hit to market share, or whether this is the start of a multi-year share shift toward vertically integrated chip strategies at the largest AI buyers.

The bear case for the selloff itself: a trillion-dollar drawdown driven partly by macro rate fears and partly by ASIC headlines that haven't yet shown up in actual revenue impact may be overpricing a real but still-early competitive shift. What to watch next: Nvidia's next quarterly earnings call for any disclosed impact from hyperscaler ASIC adoption on order volumes, and whether the stock's newly compressed multiple attracts value-oriented buyers even as growth investors rotate out.

ShareXLinkedInEmail
More onNvidia โ†’

Originally reported by TradingKey. Analysis and editorial commentary by Value Add Pulse.

โ† Back to Pulse

THE WIRE in your inboxโ€” Tech, startup & VC news with Trace's take. Free, no spam.

Read Next

AI~$1T Nvidia repriced

The AI Trade Just Split in Two: Buyers vs. Sellers

Nvidia has shed roughly $1 trillion in value since May even as Meta's data center costs balloon past $50 billion and Helsing raises Europe's biggest-ever defense round -- capex is up, but the margin is shifting away from the chip seller.

AIPT HK$1,880 (Zhipu)

Goldman Names Its Preferred Chinese AI Models

Goldman Sachs research named Zhipu, DeepSeek and ByteDance its preferred Chinese AI models, ranked ahead of Alibaba and Tencent on speed-to-market, and recommended rotating out of Korean AI stocks into a long China AI basket.

AI~$21/mo India Pro

Anthropic Localizes Claude Pricing for India

Anthropic began localizing Claude subscription pricing for India, its second-largest market at 5.8% of global usage, with Claude Pro at roughly $21/month there versus $17 in the US, though UPI payment support still isn't enabled.

@Trace_Cohenยทt@nyvp.com