Michael Burry just lit a flare over Nvidia’s $95 billion purchase-obligation bomb, telling investors the chip titan is walking the same supply-chain plank that snapped Cisco in 2000. Margins of 70 % won’t soften the fall if AI demand hiccups.
The 489 % Obligation Spike
Nvidia’s latest Form 10-K shows purchase obligations exploded to $95.2 billion from $16.1 billion in just twelve months. Add in other supply agreements and the company is now on the hook for roughly $117 billion—a liability stack larger than the market caps of Intel, AMD and Qualcomm combined. Burry’s Substack memo labels the shift “not business as usual—this is risk,” because the contracts lock Nvidia into paying for wafers years before customer demand is proven.
Cisco’s Ghost at the Fab Door
In 2000 Cisco signed long-term supplier deals to feed its projected 50 % annual growth. When telecom capex vaporized the next year, the company wrote off $2.2 billion in excess inventory and the stock collapsed 82 %. Burry sees the same blueprint: forced capacity grabs at TSMC and SK Hynix that assume infinite AI expansion. The difference—Nvidia’s 70 % net margin—is cold comfort; Cisco’s gross margin was 65 % the quarter before its cliff dive. Margins revert faster than purchase contracts can be renegotiated.
Wall Street’s Split Verdict
Rosenblatt Securities lifted its Nvidia price target to $300, arguing the commitments prove management’s confidence in next-gen Blackwell and Rubin platforms. Burry counters that the obligation surge is structural, not cyclical: “What is happening now is not temporary… it is coming from within the business plan.” The dichotomy is stark—sell-side models assume 35 % CAGR through 2030, while Burry’s scenario implies a potential $30 billion inventory writedown if hyperscaler budgets flatten.
Cash-Flow Squeeze Looming
Nvidia’s free cash flow over the last four quarters totaled $48 billion, enough to cover barely half the new commitments. Contractual wafer payments are front-loaded: 60 % due within 18 months according to footnote disclosures. If AI server builds slow, Nvidia still mails the checks, turning today’s cash cow into tomorrow’s working-capital hog. Cisco’s 2001 experience shows the street punishes that pivot with a single-digit P/E re-rating.
What History Says About Supply Traps
Semiconductor leaders have a poor record of outrunning their own purchase orders:
- Intel 1985: Over-ordered capacity ahead of the PC boom, then suffered three straight quarters of negative cash flow when the market paused.
- Micron 2007: Locked in DRAM supply before the financial crisis; stock fell 91 % as DRAM spot prices dropped 85 %.
- Texas Instruments 2012: Paid $650 million to cancel wafer agreements after smartphone growth disappointed.
In each case gross margins compressed by 20–30 percentage points within two years—far faster than equity analysts modeled.
Investor Playbook Right Now
- Hedge: Long Nvidia/short SOXX pairs trade limits sector beta if obligations sour.
- Monitor: Watch for Q2 2026 days-sales-outstanding spike—an early flag that customers are stretching payment terms as inventories build.
- Stress-test: Model a 40 % haircut to consensus 2027 free-cash-flow forecasts; the stock trades at 38× that reduced number, revealing limited downside protection.
- Track: TSMC monthly revenue reports; if wafer starts for 5 nm and 3 nm nodes flatten while Nvidia obligations rise, the divergence signals oversupply.
Red Flags That Confirm Burry
Three metrics now flash yellow simultaneously:
- Purchase obligations / TTM revenue hit 1.8×, a record high versus a five-year median of 0.6×.
- Inventory on Nvidia’s balance sheet rose 43 % faster than sales last quarter, the first divergence since 2018.
- Hyperscaler capex guidance for 2026 sums to $215 billion, flat versus 2025, yet supplier obligations imply 40 % unit growth.
The math only works if enterprise and sovereign AI demand magically backfills, a bet Cisco also took in 2000.
Stay ahead of chip-sector implosions and the next big short—bookmark onlytrustedinfo.com for the fastest, most authoritative analysis before the market opens.
