AI search startups surge in funding as Google remakes search
Exa raises $250M at a $2.5B valuation, Parallel adds $100M — and Alibaba counters with a 3x-faster AI chip.
One-Line Summary
Money and interfaces move in tandem: AI search startups attract big funding as Google shifts search toward AI answers, while Alibaba strengthens the compute stack with a faster homegrown chip and a new model.
Big Tech
Alibaba unveils 3x-faster Zhenwu AI chip, teases next Qwen model
Alibaba, the China-based e-commerce and cloud provider, introduces a new AI processor and previews a next-generation model, signaling more in-house infrastructure for AI workloads. The Zhenwu M890 is described as delivering three times the performance of the Zhenwu 810E, with 144 GB of GPU memory and 800 GB/s interchip bandwidth; Alibaba also says its next-generation large language model, Qwen3.7-Max, will be released soon. 1
Alibaba says it has delivered 560,000 Zhenwu units to more than 400 customers across 20 industries, and earlier in April it announced a China data center with China Telecom powered by its own chips — positioning homegrown compute as Nvidia faces constraints shipping advanced chips into China. For teams serving users in China, this points to more local options for training and inference. 1
Industry & Biz
AI search startups win big funding as Google reworks search
Startups building AI-first ways to find information online are raising large rounds to challenge traditional search results pages. TechCrunch reports that Exa Labs raises $250 million at a $2.5 billion valuation (citing Bloomberg), alongside peers like Tavily, TinyFish, and Parallel Web Systems; Parallel, led by former Twitter CEO Parag Agrawal, reportedly raises $100 million at a $2 billion valuation (citing the Wall Street Journal). 2
TechCrunch also notes that Amazon, LinkedIn, and Reddit are revamping search and discovery with AI — creating likely acquirers if a startup breaks out. It characterizes ChatGPT as owning the “interface layer” and says it handled the vast majority of AI-powered searches before Google’s AI launch, while OpenAI does not prioritize search and Google must consider ads — leaving room for focused entrants to find niches. 2
For business teams, this funding wave signals a distribution shift: more answers will come from AI summary boxes across multiple surfaces, not just 10 blue links. Watch how these startups source and index the web, what they charge, and whether incumbents move to partner or acquire — TechCrunch points to multiple potential buyers across Amazon, LinkedIn, and Reddit. 2
What This Means for You
Traffic may increasingly arrive through AI-generated answers across different apps, not only from classic search results. Map your top queries and observe which sources AI systems cite — presence there increasingly determines discovery. 2
Restructure content for “answerability.” Short, factual summaries with clear headings, Q&A blocks, and bullet points improve the odds of being quoted inside AI search experiences highlighted by TechCrunch. Keep product pages and FAQs current so both startups and incumbents can cite them cleanly. 2
Support and product documentation may need to be written more like reference facts than marketing copy so AI systems can quote it precisely. Treat AI answer surfaces as another distribution channel for help content and comparisons. 2
If you operate in or sell to China, Alibaba’s 3x-faster Zhenwu M890 and the forthcoming Qwen3.7-Max point to expanding local compute and model options. Ask local partners how availability and pricing could affect your stack and data residency plans. 1
Action Items
- Run an AI-answer visibility check: List your top 10 queries and test them in ChatGPT and Google’s AI experiences; note which sources and pages get cited.
- Make one page “answer-ready”: Rewrite a key FAQ or product page into a 150–300 word, bullet-heavy explainer with clear headings and concrete facts.
- Create a citations tracker: Start a simple spreadsheet logging where AI answers pull from for your brand and competitors; revisit weekly to spot shifts.
- Scope a China-market inquiry: If relevant, ask your local cloud or systems integrator about timelines and pricing for Alibaba’s M890 and Qwen3.7-Max to understand options.
Comments (0)