DeepSeek considers doubling latest funding round to up to $15 billion, sources say

DeepSeek is considering increasing its funding round target from 50 billion yuan to as much as 100 billion yuan ($14.9 billion), while seeking a valuation of about 500 billion yuan ($75 billion). The round has attracted state-backed and corporate investors, including reported commitments from CATL and Geely Auto, but has taken longer than planned amid investor concerns over price; the fundraising remains fluid. DeepSeek has also tapped CITIC Securities to prepare for a possible Shanghai STAR Market IPO, with an early-2027 listing previously reported.

Analysis

The funding is a signal about capital allocation, not yet proof of monetizable AI demand. If the new money supports larger training runs and inference capacity, it could lift near-term demand for advanced compute—including a potential second-order benefit to NVIDIA (NVDA) if DeepSeek continues relying on its chips. The opposing 6–18 month risk is that more capital accelerates model efficiency and domestic-chip substitution, weakening compute intensity per unit of capability and eventually limiting NVDA’s China-linked opportunity. The key variable is hardware mix, which this report does not establish.

The implied step-up in private valuation, alongside reported investor price resistance and a possible IPO pipeline, raises the bar for proof of commercial returns. State and corporate investors may tolerate strategic objectives that do not maximize near-term profit; that could sustain spending, but makes private-round pricing a poor read-through to public-company earnings. For CATL and Geely Auto, absent disclosed strategic or financial terms, treat participation as a capital-allocation and governance watch item—not evidence of material operating synergies.

Near term, funding headlines may support AI sentiment and NVDA, but the information is sourced to people familiar with private talks and the round remains fluid. Over 1–3 months, verify final terms, investor commitments, and evidence of compute procurement. The structural risk to the bullish NVDA read-through is a shift toward non-NVIDIA hardware or falling compute requirements; evidence of continued NVIDIA dependence would falsify that concern.

Key Decisions for Investors

• Do not trade NVDA solely on the headline. Treat any near-term AI-sentiment lift as tactical; look for confirmed funding closure and evidence of incremental compute purchasing before adding exposure.
• Set a 1–3 month diligence alert for DeepSeek hardware disclosures, procurement signals, or credible reporting on chip mix. Continued reliance on NVIDIA supports the demand thesis; meaningful domestic-chip substitution weakens it.
• Avoid using the reported private valuation as a public-market comparable or as proof of earnings power. Reassess only when the financing terms, dilution, and commercial monetization evidence are clearer.
• Monitor CATL and Geely Auto for disclosed investment size, strategic rationale, and materiality relative to their businesses. Without those details, no company-specific trade is warranted.