Mamoon Hamid on AI Investing, Founder Conviction and Venture Returns

Kleiner Perkins investor Mamoon Hamid joins Harry Stebbings to discuss AI investment, venture-fund construction, founder selection and liquidity. Hamid's firm has invested in several companies he praises, and he says he is a limited partner in Stebbings's funds; their comments reflect investment interests.

Hamid sees AI as a major technology cycle, comparing the excitement to the internet era he experienced after arriving in Silicon Valley in 1997. He expects substantial value creation but acknowledges that large incumbents can spend enormous sums on infrastructure and models. The scale and distribution of future value are forecasts.

Kleiner has backed applications aimed at helping highly skilled workers, including doctors, lawyers and developers. Hamid names Ambience, Harvey and Codeium. He sees opportunities in solving specific workflow problems, while noting that his fund has financial interests in those companies.

In crowded categories, he emphasizes technical quality, domain knowledge and execution. His assertion that a medical transcription product needs close to 99% quality is an illustrative standard, not a measured performance claim or a substitute for clinical validation.

He says the basic venture task remains finding strong founders in the right market at the right time, even when AI changes quickly. Stebbings asks how to navigate unusually high pre-product valuations; Hamid says exceptional founders may justify occasional exceptions, but a fund cannot make every investment that way.

For early-stage economics, Hamid describes investing $5–10 million for roughly 15–20% ownership as an illustrative target for his funds. A much higher valuation can make the required return harder. He says the firm keeps an exception bucket for unusually strong conviction rather than treating every exception as the strategy.

He argues that AI products may capture some value associated with labor, not just traditional per-seat software budgets. His example compares $30 and $300 monthly prices across 1,000 seats. That is hypothetical arithmetic about pricing, not evidence that any specific product reliably delivers a five- or tenfold productivity gain.

On Klarna's proposal to replace some SaaS with internal tools, Hamid recalls that an internally built CRM at Khosla Ventures cost millions to build and maintain. He understands why a company might use developers and AI to automate work, but thinks many will ultimately buy a specialized product and pay for outcomes.

Many large companies are spending on AI proofs of concept, he says, and some will discover that maintaining custom systems is harder than the demo. He compares this experimentation to early internet consulting. His expectation that automated customer support will eventually improve the experience is a prediction, not a current guarantee.

Hamid worries that some middleware between foundation models and applications may be overfunded when technology changes rapidly. He expects opportunities at the model, infrastructure and application layers, though he acknowledges uncertainty about how durable any individual layer's value will be.

Asked whether foundation-model providers will absorb applications, he says they cannot build every vertical product and sees room for specialized companies. He recounts hearing that view during an OpenAI visit. This is his investment thesis, not a binding commitment by model providers to stay out of applications.

Stebbings questions the economics of models as prices fall. Hamid distinguishes chip suppliers, cloud infrastructure and model providers; he thinks providers may eventually find profitable margins but does not establish that all model businesses are profitable today.

He estimates token prices had fallen roughly 200-fold over the prior 18 months and speculates about a further 10- to 20-fold reduction over two years. Comparability depends on model, task and pricing; the first is his rough estimate and the second a forecast, not a demonstrated scaling law.

To address concerns about AI infrastructure spending, he sketches a scenario where technology captures a larger share of a growing global economy, creating additional annual demand. The GDP, labor and technology shares are his rough assumptions, and the projected spending does not establish returns on any particular infrastructure investment.

What remains constant in venture, he says, is backing founders solving important problems. What has changed is the amount of available capital: multiple competitors can be funded heavily at once, and some may be overfunded. He considers this both opportunity and risk.

Hamid describes Kleiner as early-stage focused, with an $800 million early-stage fund and a $1.2 billion growth fund managed by a small team. He says roughly half of growth-fund dollars go to existing early-stage portfolio companies. These are fund structures described at the interview, not current holdings.

For reserves, he gives an illustrative $25 million lifetime investment: $15 million initially and $10 million for later rounds, roughly a 60/40 split. Funds are reallocated as companies fail, exit or need different amounts. He says the firm often participates at least modestly in follow-ons to manage signaling, while acknowledging different circumstances.

A pay-to-play question brings up Box. Hamid recalls three bridge financings around 2008–09 when investors were reluctant during the financial crisis. He says investors backed Aaron Levie and Dylan Smith and invested further at a roughly $25 million valuation; the company's later public listing shows survival, not that bridge financing always succeeds.

He regrets the dilution Box's founder experienced through repeated bridges and says he has apologized to Levie. He believes investors and boards should pay attention to founder ownership and replenish incentives when appropriate, while recognizing that rescue capital itself has a cost.

Companies plateau when they stop innovating and concentrate on defending existing territory, in his view. Selling is also difficult: Hamid looks for a local high in perceived value, especially when a strategic buyer may build more with the product. He acknowledges that correctly identifying that point is hard.

He cites Microsoft's $1.2 billion acquisition of Yammer as an exit that returned cash and later left his firm free to invest in Slack. He says distributing public stock can let limited partners decide whether to keep holding, even though a sale or distribution may sacrifice later upside.

Hamid calls Slack one of his strongest multiple outcomes and mentions Figma and Rippling at their initial post-money valuations. He also describes a $3 million investment in a company later sold for $700 million, illustrating the effect of time on annualized return. Some cited company values are private marks, not realized proceeds.

His numerical illustration of $3 million becoming roughly $170 million over 15 years is not consistent with the approximately 15% annualized return he quotes: that simplified gross calculation is about 31% annually before fees and other adjustments. The broader point—that holding period changes annualized return—still holds; the interview does not supply enough detail to calculate the actual fund return.

Hamid sees a slow acquisition market and says large buyers are cautious about regulatory scrutiny. Stebbings observes that private companies can buy smaller businesses with stock whose value is difficult to assess. Hamid notes that concerns around the abandoned Adobe–Figma deal involved regulators beyond the US FTC, based on his experience as an investor.

He says the IPO market was quiet before the election and hopes the following year will improve. That is a prediction made at the time of the interview, not a confirmed reopening or a current market assessment.

Figma attracted him before significant revenue because designers used it intensely. He recalls early cohorts using the collaborative browser product 15–18 days a month, despite relatively few users and only hundreds of thousands of dollars in revenue. He credits founders Dylan Field and Evan Wallace and describes his firm's investment interest.

His investment thesis included possible expansion from designers to adjacent product, marketing and engineering users. Hamid says competitive markets can be attractive, but especially likes products that create a new category; he cites Slack, Figma and portfolio company Glean as examples of his view.

He describes two founder profiles he likes: first-time founders deeply obsessed with a product and emerging market, and repeat founders determined to exceed their earlier work. He is less persuaded by a purely top-down market-sizing exercise without that personal connection to the problem.

Experience can justify a premium, but price still matters. He says Kleiner and Lightspeed backed Glean at a roughly $35 million post-money valuation, and contrasts it with a hypothetical much higher price he would probably decline. His view reflects an investor's incentives as well as a concern about later financing.

Hamid disagrees that founders should always raise the maximum at the highest valuation. He prioritizes choosing partners and having enough cash; if a company already has substantial reserves, he says, its leaders may be better off building. Stebbings adds that an excessively high valuation can complicate the next round.

He is reluctant to back even a great founder in a structurally poor market with compressed margins or difficult customers. Stebbings argues a strong founder can pivot to a better market. They leave a genuine difference in emphasis rather than a formula for every company.

Slack was not an obvious home run, Hamid says. He recalls investing at about $250 million post-money when ARR was around $500,000 because a meaningful fraction of roughly 10,000 users spent hours in the product daily. He argues that engagement and potential breadth can be more informative than a revenue multiple at that early stage.

He is skeptical that proprietary data platforms alone provide a decisive advantage in early-stage investing. Access to information matters, but founders must also choose the investor. Stebbings says Hamid is a limited partner in his funds, and both discuss the importance of reputation and prior work.

Kleiner's partners discuss deals rather than holding a formal vote, Hamid says. Questions about the market, competition or price can weaken a lead partner's conviction. He says a partner could veto a deal, but none had done so; discussion and veto are therefore distinct from a simple majority vote.

He has moved away from assigning exact probabilities to many exit scenarios because he sees false precision. Partners with similar instincts can reinforce one another but may also miss the same things, he acknowledges. Figma itself prompted considerable questions within the firm before the investment.

Hamid says a fund raised by the newer team was deployed in about 15 months, faster than the preferred two-and-a-half to three years. It included Rippling and Glean, and he expects it to perform well. That expectation is not a reported realized fund return, and faster deployment reduced diversification across time.

He describes Tally's shutdown as a loss of roughly $30 million, including follow-on capital. He says rising interest rates made consumer lending difficult. His lesson is that lending is hard and that additional investment can magnify losses even when earlier rounds attracted respected investors.

Hamid tries to remain open-minded despite experience, but accepts criticism that his optimism about founders can persist too long. He believes good investor advice can help and poor advice can destroy value; board meetings work best, he says, when they address one or two consequential issues deeply.

He prefers capital-efficient businesses and says he does not want Kleiner to become a capital accumulator. The firm's stated approach is to find a small number of exceptional early investments and selectively back the strongest later; this is a strategy, not a promise of results.

He recalls meeting Box founder Aaron Levie in 2007 and recognizing how deeply he had thought about cloud storage. In quick fire he says a return to more normal historical public-market valuations would help venture, names investors he respects and says he would be curious to lead OpenAI for a day.

Hamid identifies geopolitical tension and polarization as concerns. Asked what question he is rarely asked, he discusses faith as a daily influence on humility, empathy and how he treats people and the planet. Stebbings closes with a joke about liquidity, and both thank each other.