Menlo Ventures Partner Deedy Das: Multiple Seed Investments are Options
Menlo Ventures partner Deedy Das responded to the claim that "seed investments are options," stating that if a venture capitalist mocks this framework, their entire investment should ideally be just one check; otherwise, by definition, they are buying options for themselves.
His argument directly targets portfolio behavior. As long as a fund holds more than one company, each early check is purchasing the right to observe outliers, rather than making a one-time belief bet. Those who mock option thinking, if they simultaneously diversify across multiple companies, are logically executing the same strategy.
Das was promoted to partner at Menlo in September 2025. Before joining, he was a founding team member at Glean, progressing from engineer to engineering manager, and then overseeing the Glean Assistant product. During his tenure, the company grew from fewer than 10 people to over 300, achieving a valuation of $2.2 billion from having no product. Earlier, he worked in technology at Facebook and Google, participating in search and AI query understanding at Google.
At Menlo, he was involved in launching the $100 million Anthology Fund in collaboration with Anthropic, which has invested in over 40 companies in the early stages of AI. Public investments include Goodfire, where he serves as a board member. He initially entered OpenRouter with a small check from the Anthology Fund, later leading a $40 million Series A round in June and serving as a board observer.
OpenRouter provides developers with a unified interface to access hundreds of models from Anthropic, Google, Meta, Mistral, and others. Das noted that in the past year, its annual token processing increased from about 100 trillion to over 250 trillion, approximately 25 times. Since its revenue is tied to usage, the surge in usage corresponds to a simultaneous rise in revenue. Its public portfolio also includes Wispr Flow, Modal, Pangram, Inception, and Prime Intellect.
This is an alignment of positions within the same institution, not a new fundraising event. Buyers treat early checks as information options and retain the right to increase their stakes later. The pressure on sellers lies in being limited to a single small proportion and unable to expand their positions after data confirmation as seed managers. Mockery itself does not change the flow of funds; multiple holdings are, by definition, an options portfolio.
Menlo's website lists its responsibilities as early investments in AI infrastructure, robotics and hardware, and enterprise software. The structure of the Anthology Fund is designed to exchange access to early-stage projects for relationships with Anthropic's models, with the main fund increasing its stake once evidence of usage or revenue appears.
Source: Public Information
ABAB AI Insight
Das's background is more product-oriented than purely financial. He worked on query understanding at Google Search, then helped build Glean's assistant product into a significant revenue component, with the company's valuation rising from no product to $2.2 billion, and subsequently reported to have reached $7.2 billion. This trajectory influences his approach to early projects, where he first looks at whether usage can translate into revenue, rather than focusing on ownership percentages. His promotion to partner occurred about a year and a half after joining, with Menlo publicly stating that partner promotions happen only once every few years.
The movement of money occurs on two levels. The first level is the $100 million Anthology Fund in collaboration with Anthropic, which has covered over 40 early-stage companies, functioning as a low single-check, high-quantity observation fund. The second level is the main fund exercising its options: OpenRouter first received a small check, and after usage expanded from about 100 trillion tokens to over 250 trillion, Menlo led a $40 million Series A round. Goodfire directly secured a board seat. The motivation is to exchange distribution relationships with model vendors for project access, then concentrate confirmed capital on names with steep usage curves.
This mirrors Glean's own early experience of being entered by multiple funds with small amounts, then scaling up valuation after product validation, and is similar to the strategy in enterprise software during the 2010s of first securing design collaborations, then increasing stakes after annual recurring revenue appears. The current position is not a survival argument for seed funds, but rather a formal definition of early portfolios as options pools by platform funds, with the industry in a control phase after expansion: access relies on model relationships, and returns depend on who can issue the next check when 25 times usage appears.
Structurally, this represents a transfer of pricing power. Early prices are no longer used to prove correct judgment but to purchase subsequent allocations. The mechanism is that model calls have tied the costs and revenues of startups to usage, allowing revenue from companies like unified gateways to be directly verified by token counts. Once verification metrics are public, the choice for excess disappears, and platforms capable of continuous funding gain pricing power, while funds that remain at the first check are left with diluted observation positions.
ABAB News · Cognitive Laws
Holding multiple companies is, by definition, buying options.
Those who mock diversification often diversify themselves.
The first check buys eligibility, the second check buys returns.