DeepSeek raised $7B in its first major funding round, abandoning its earlier 'no funding, no commercialization' pledge. The Chinese AI lab plans to double headcount across all departments and launch a Harness AI coding agent team to compete with Anthropic's Claude Code.
Key facts
- DeepSeek raised $7B in its first major funding round.
- Company plans to double headcount across all departments.
- Launches Harness AI coding agent team to compete with Claude Code.
- AI startup Lindy ditched Claude entirely for DeepSeek due to cost.
- Alibaba allegedly used 25,000 accounts to mine Claude over 28.8M exchanges.
DeepSeek raised $7B in its first major funding round, abandoning its earlier 'no funding, no commercialization' pledge. The Chinese AI lab plans to double headcount across all departments and launch a Harness AI coding agent team to compete with Anthropic's Claude Code According to Pandaily.
The pivot comes as cost pressure mounts on US AI labs: AI startup Lindy ditched Claude entirely for DeepSeek, with CEO Flo Crivello calling it 'a matter of survival for the business' [The Decoder reports]. Lindy's AI costs had exceeded personnel costs, mirroring a broader trend as Chinese models undercut US pricing.
DeepSeek's expansion coincides with aggressive moves by other Chinese players. Alibaba allegedly used 25,000 accounts to mine Claude over 28.8 million exchanges, per Ars Technica, signaling a data-scraping arms race. Meanwhile, new models launching in Asia promise Mythos-like capabilities without export-ban risk, threatening to permanently lock US AI labs out of the region [TechCrunch notes].
The US government approved Anthropic's limited re-release of Claude Mythos 5 to organizations running critical infrastructure, with broader access still under negotiation [NBC News reports]. The approval comes after Anthropic voluntarily suspended Mythos under regulatory pressure in late June, per The Decoder.
DeepSeek's $7B round—its first external capital—represents a strategic reversal from its idealistic roots. The company did not disclose the valuation or lead investors. With headcount set to double and a dedicated coding agent team in the works, DeepSeek is signaling it intends to compete head-on with Anthropic's Claude Code, which scored 69.2% on SWE-bench Pro with Opus 4.8.
The funding arms race is intensifying as Anthropic targets a 2026 IPO at a $1T+ valuation [per recent history]. DeepSeek's cash injection gives it runway to undercut pricing and capture price-sensitive developers—a segment where US labs are vulnerable.
What to watch
Watch for DeepSeek's Harness AI coding agent benchmarks on SWE-bench and Terminal-Bench, and whether the company discloses its valuation or lead investors. Also track Anthropic's Q3 ARR disclosure and enterprise seat counts for Claude Code, which will reveal whether cost pressure is eroding market share.
Source: pandaily.com
[Updated 28 Jun via pandaily]
DeepSeek founder Liang Wenfeng published his first post-funding research paper, introducing the DSpark speculative decoding framework that boosts generation speed by 85%, signaling the company is investing in technical differentiation alongside its commercial push [per Pandaily].
