Evidence by mid‑2025 shows multiple AI unicorns in both of the categories Sacks described.

1. AI infrastructure unicorns ("vector databases, agent platforms," broader infra)

• CoreWeave: An AI cloud‑infrastructure provider focused on GPU‑based compute for AI workloads. By May 2024 it raised a $1.1B round at about a $19B valuation, and its IPO in March 2025 valued it around $19–23B, clearly making it an AI infrastructure unicorn well before and by mid‑2025. (time.com)
• Scale AI: Provides data and evaluation infrastructure for training and deploying AI models. It has been a unicorn since 2019 and reached $13–14B valuation in 2024, explicitly framed as core AI infrastructure. (cxotech.com)
• Crusoe: Described as a “vertically integrated AI infrastructure provider” with AI‑optimized data centers. It joined the unicorn club in 2022 and later raised funding at a $2.8B valuation in 2024 and $10B valuation in 2025, confirming its status as an AI infrastructure unicorn throughout the relevant period. (investclub.sv)
• 01.AI: An LLM company that developed efficient AI infrastructure and inference engines; it achieved unicorn status (> $1B valuation) in November 2023, again prior to mid‑2025. (en.wikipedia.org)

These examples show far more than two distinct AI infrastructure unicorns existing by mid‑2025, in line with Sacks’s expectation for that segment.

2. Vertical/professional AI copilot unicorns

• Poolside: Builds an AI‑powered coding copilot; in June 2024 it raised $400M at a $2B valuation, making it a unicorn developer‑tools copilot. (interstellardesign.studio)
• Harvey: Markets itself explicitly as an AI "copilot" for lawyers, automating legal research and drafting. A July 24, 2024 funding round valued Harvey at $1.5B, establishing it as a legal‑AI copilot unicorn well before mid‑2025; subsequent rounds later pushed this much higher. (news.aibase.com)
• OpenEvidence: Business Insider’s May 7, 2025 report describes OpenEvidence as an AI copilot for doctors that helps clinicians access and interpret medical research, noting it raised capital at a $1B valuation with only ~20 employees. (businessinsider.com)

By mid‑2025 there are therefore multiple unicorns each for coding, law, and medicine copilots, satisfying the prediction that there would be “multiple unicorns” in the professional copilot category.

3. Overall context: “Dozens of new unicorns” in AI Independent market analyses report dozens of new AI unicorns: one 2025 funding overview notes 47 new AI unicorns in 2024 alone, across infrastructure, healthcare, platforms, and applications. (secondtalent.com) This supports Sacks’s broader claim that many unicorns would be minted across AI, not just in the two example sub‑segments.

Given (a) the clear presence of multiple AI infrastructure unicorns and (b) multiple profession‑specific copilot unicorns by around mid‑2025, Sacks’s prediction—both in spirit and in the normalized form you provided—has been borne out by events.