Mitchell Green on Software Stocks, Venture Liquidity and Investing Through AI Change

Harry Stebbings asks Mitchell Green, founder and managing partner of Lead Edge Capital, whether the software-stock sell-off reflects a justified reassessment or an overreaction to AI. Green says his firm is buying public software shares, including Procore, Workday and Appian, and has repurchased Toast after an earlier investment and sale. Those positions are relevant to his assessment of the sector.

Green argues that incumbents have distribution, data and balance sheets that can help them adapt. He does not think every incumbent will survive: in his view, technological disruption creates new companies, lets some existing ones thrive and can destroy others. His description is an investment thesis, not a guarantee for any named business.

Stebbings questions the outlook for Workday given slower growth, seat-based pricing and AI competition. Green says Workday has a rapidly growing AI business and substantial revenue and cash flow, although the figures he cites are his own estimates. He argues that the scale of an established company makes its growth profile different from that of a fast-growing private AI lab.

Green thinks some software analysts had not adequately accounted for slowing growth as companies became larger. He predicts estimates will be reduced and that companies could later beat the revised expectations, though he expects some stocks may remain stagnant in the near term. This is his interpretation and forecast, not a demonstrated explanation of every stock decline.

Stebbings invokes the difficulty of buying a falling stock and points to Duolingo as a position where the decline continued. Green describes buying gradually or on down days as one way he thinks about timing, while acknowledging that investors cannot reliably identify the bottom. His discussion is a personal approach, not individualized investment advice.

Asked whether non-founder CEOs are inherently disadvantaged during an AI transition, Green partly agrees that entrepreneurial leaders can be valuable but says capable professional CEOs also exist. He looks for a management team oriented toward growth and innovation. In his view, a company managed primarily for near-term margins may have less room to respond to a major technology shift.

Green identifies heavy leverage as a particular risk because interest obligations can constrain investment. He compares traditional retailers that adapted to e-commerce with others that failed, attributing some of the difference to financial flexibility. Those historical examples are his interpretation; debt alone does not establish why a particular retailer succeeded or failed.

The guests debate how much major technology companies should spend on AI infrastructure. Green says the right answer between restraint and very heavy investment is unclear and may lie somewhere in between. He argues that Meta’s leadership has reasons to invest aggressively when competitors are doing the same, while acknowledging that only time will show which strategy works.

Green calls ByteDance the most advanced AI company in the world and says Western observers underestimate its use of and investment in AI. Lead Edge has invested in ByteDance, an interest relevant to that judgment. His ranking is an opinion, not a comparative study establishing that the company leads every AI field.

He expects AI to create major companies beyond the first wave of familiar applications. Drawing an analogy with social media businesses that emerged after the early internet era, he thinks some of the largest future businesses may begin in the next two to five years. He says he does not know what those businesses will be; this is a prediction, not a timeline for particular outcomes.

Stebbings asks whether investors should keep investing or wait for a new equilibrium. Green says the answer depends on the fund’s strategy: an early-stage investor seeking rare 100-fold outcomes has different economics from his own growth fund. He likes to ask whether a company could justify the entry price within about 18 months if it meets a reasonable operating plan.

Green describes Lead Edge’s target as roughly two to five times invested capital over three to seven years, aiming for approximately 25% internal rate of return on its investment curve and around two to two-and-a-half times net at the fund level. These are stated targets and his account of the strategy, not audited guarantees of future returns.

He says investing feels harder than in 2017 but sees opportunities in secondary and special-situations transactions. He cites a recent deal where a new financing was being discussed at roughly twice Lead Edge’s entry price. The example is his account of an interim valuation, not a realized return or independent verification of the later round.

Green points to Grafana Labs as a portfolio company he believes benefits from AI-related spending, saying Lead Edge and another investor backed it when it was a much smaller bootstrapped software business. He says his team found the company through persistent cold calls. His firm’s investment and his account of company growth are material context for his enthusiasm.

He describes Pacemate, another portfolio company, as software that brings cardiac-device information together for clinics. Green cites its revenue growth and approximately 99% gross dollar retention as company figures, and says AI could help support work become more productive. These are his business claims, not evidence of a clinical benefit or a substitute for medical oversight.

Green argues that AI may improve productivity in sales, marketing and customer support as well as engineering. He estimates that research and development represents about 30% of cumulative spending for an average software company at IPO, using that figure to explain why distribution and service matter. The percentage is a rule of thumb from the interview, not a verified industry census.

The guests discuss a widely circulated research piece and the speed at which social media can move investor attention. Green contrasts recent volatility with his experience at a hedge fund during 2008 and 2009, when he recalls far larger market swings. He sees fear-driven price movements as potential opportunities, while Stebbings worries that prices may become detached from underlying value.

Stebbings argues that today’s AI shift differs from a financial crisis because new technology could displace incumbents. Green responds that mainframes and longstanding companies such as Oracle, Microsoft and SAP have persisted through earlier changes. He expects both survivors and failures, and favors businesses with strong gross revenue retention; these are investment views, not a forecast for each company.

Green speculates that AI could speed drug development and eventually help address diseases such as cancer and dementia. The interview does not establish that AI has solved those conditions, validated faster trials or produced a particular treatment. He also expects unlevered manufacturers may have more freedom to invest in robotics than heavily indebted competitors.

Stebbings asks whether productivity gains could weaken consumer spending if jobs disappear. Green says he is less worried about mass unemployment, citing earlier technological transitions and a conversation with a senior bank executive who anticipated retraining workers. One executive’s account is not evidence of what every employer will do, and Green’s employment outlook is a prediction.

Green argues that adoption at large regulated companies may be slower than technology observers expect. He expects new roles and businesses to emerge and does not believe the most extreme unemployment scenarios will materialize. His historical analogy does not rule out job losses or establish how quickly workers could transition.

The conversation returns to social media and market behavior. Green says information now moves very quickly and expects greater attention to regulation, while acknowledging that market structure also includes larger passive funds. He favors examining earnings and free cash flow, but the existence of either does not guarantee a share-price floor or a profitable investment.

Green raises stock-based compensation and dilution as costs he believes investors sometimes overlook. More shares can dilute existing owners even when reported operating figures look attractive. He contrasts that with repurchases, while noting that his firm is not an activist investor; whether a buyback creates value depends on its price and financing.

Discussing insider purchases and company buybacks, Green says he would consider them one positive signal but asks how much stock an executive already owns. He expects more repurchase programs from some companies. The panel’s references to particular planned amounts and individual purchases are conversational figures, not verified transaction records.

Stebbings asks how a valuation discount for ByteDance related to China could narrow. Green compares earnings multiples of Chinese public companies and a Western peer, then argues that ByteDance’s growth and earnings could support a higher valuation. He discloses that Lead Edge owns substantial ByteDance stock; his multiples and estimates are time-specific and unverified in the interview.

Green speculates that ByteDance could eventually list in Hong Kong and reach tens of billions of dollars in annual earnings within five years. He immediately acknowledges uncertainty over the listing question. Neither an IPO venue nor the projected earnings are established, and his earlier phrasing about US listing should be understood in light of that uncertainty.

He says Lead Edge bought ByteDance shares at valuations around $200 billion and views the risk-adjusted opportunity favorably. He predicts that China could lead in AI, citing energy infrastructure, technical talent and scientific resources, while acknowledging that competition is not necessarily winner-take-all. These are his investment position and geopolitical forecast, not a measured outcome.

Green asks whether leading model companies can ultimately earn durable profits and says he does not know. He also predicts community resistance to data centers where local residents perceive higher power costs or environmental effects without sufficient local benefits. His examples of local prices and employment are illustrative scenarios, not verified findings for specific communities.

Asked about selling, Green summarizes his discipline as continually re-underwriting each holding. The question is whether the current valuation still offers an attractive probability of the targeted return, not just whether the company was a good investment at the original price. Lead Edge’s stated goal remains two to five times capital over three to seven years.

Using ByteDance as an example, Green discusses reported secondary prices and says his willingness to sell would change at a much higher offered valuation. He also forecasts substantially higher future earnings. The prices and earnings are reported or hypothetical figures from an investor with a position, not proof of completed transactions or future results.

He explains the 18-month test with a company at $20 million in revenue: if it meets a reasonable plan, could a sensible valuation support the investment price after 18 months? Green distinguishes a good company from a good investment and says even an excellent business can produce a poor return when bought too expensively.

Green describes structured secondaries and purchases of old fund interests as potential ways to buy quality assets at better prices. He says some investors seek extreme outcomes and accept very different pricing. His approach is not that weaker companies are automatically attractive; it is that business quality and entry price have to be considered together.

Green says he believes roughly half of venture investors, perhaps more, should not be in the business because too much capital is chasing opportunities without adequate price discipline. The percentage is his provocative opinion, not a measured count. He also cautions that only a small number of companies reach the largest valuations.

He predicts that venture capital may have less money in several years, but admits uncertainty about the timing. His advice to emerging managers is to use liquidity windows, including partial sales, rather than assume paper marks will turn into cash automatically. A partial sale can help return capital while preserving some exposure, although timing and access vary by deal.

Green summarizes the distinction as “marks are opinions; DPI is math”: distributions to paid-in capital reflect money returned relative to contributed capital, whereas a private valuation is an estimate. He says about a third of his firm’s deals have involved secondary sales and emphasizes that limited partners are clients who expect capital back. Those figures are his description of Lead Edge’s history.

The guests note that a smaller fund may be able to sell a modest stake with less market signaling than a major shareholder. Green says an investor who has supported a founder can continue helping even after selling some or all of a position. This is his account of how his firm manages relationships, not a guarantee that every founder will welcome a sale.

Green estimates that a substantial portion of venture investors may add negative value, citing pressure to burn cash, poor recruiting advice and acting as if they have run a company. He explicitly says he has not run an operating company himself. His percentage is an opinion; his proposed remedy is to connect founders with experienced operators and help them recruit.

Stebbings suggests that investors push growth because their entry prices require a rapid next financing. Green agrees that high prices can create pressure and criticizes very high valuations for early ideas. When they discuss Anthropic’s earliest financing, both express uncertainty about the actual valuation; the interview does not establish those terms.

For software founders, Green identifies gross dollar retention as a central metric. His example starts with $20 million of revenue from an existing customer cohort and measures how much remains a year later after losses and reductions, excluding upsells. In his framework, $18 million is 90% retention, $19 million is 95%, and $19.5 million is 97.5%, approximately 98%.

Green says low gross retention can force a company to spend heavily on sales simply to replace lost revenue, whereas strong retention supports efficient growth. His 90%, 95% and roughly 98% thresholds are Lead Edge’s stated screening preferences, not universal standards or a complete measure of a company’s health.

Asked about private-equity-owned software companies, Green distinguishes growth investors from leveraged buyout funds. He says he has not studied the finances of the specific companies Stebbings names and therefore does not know exactly how their owners improved margins. He hopes they reduced inefficient overhead without undermining engineering, but presents that as speculation.

Green returns to the risk of debt during a technology transition: high interest payments can leave less cash to invest in product changes. Stebbings worries volatile public markets could persuade strong companies to stay private longer, worsening liquidity for fund investors. Green agrees that limited partners may eventually pressure funds to produce distributions.

Green describes Lead Edge’s secondary investment in EquipmentShare, an equipment-rental business that uses technology, and says public listing can improve customer confidence and brand recognition. His firm’s stake is relevant to the example. He also understands why founders with ample cash may prefer to stay private despite investors’ need for liquidity.

Lead Edge prefers founders who are open to an eventual public listing when the business is suited to it, Green says, because the firm needs a path to exit. He acknowledges that secondary markets can provide liquidity as well. A possible listing is a preference and potential route, not an assurance that any company will go public.

The guests discuss a hypothetical Stripe–PayPal transaction. Green explicitly says he does not know whether Stripe is pursuing it, then explains how publicly traded stock could serve as consideration in a merger. Their valuation and financing examples are hypothetical and should not be read as confirmation of negotiations or a proposed transaction.

Stebbings asks about differences between public and private company valuations. Green says private financing prices can look disconnected from public-market comparisons, and argues that some growth investors could benefit from the flexibility to buy public shares. He distinguishes that approach from an early-stage venture mandate, where the investment and risk profile differ.

On public-company size, Green says a $5 billion to $10 billion listing can be viable and that smaller companies can also list, even if attracting investor attention is harder. He points to historical examples as illustrations. His preference is an opinion about market access, not a guaranteed valuation or liquidity outcome for a new issuer.

Green questions whether very large venture and growth funds can generate strong returns without owning stakes in exceptionally valuable companies. Stebbings argues that the ability to invest billions in an eventual winner could change the math. Green agrees that large funds could find major AI winners but believes fund size and entry price make the required outcomes difficult.

He rejects the idea that firms must be either enormous or boutique to succeed, citing Benchmark and Index as examples of funds he respects. He notes that smaller funds may invest earlier and accept dilution over time. He says Index’s long-term distributions and returns impress him; those are his assessments rather than independently audited performance figures.

On whether prominent backers can make a market winner simply by supplying capital and reputation, Green says abundant capital can also fund several competitors. He praises Menlo’s early Anthropic investment as a bet that was not obviously correct at the time, and discloses a personal friendship with Matt Murphy. The final outcome of any investment remains subject to future events.

In the quick-fire round, Green says he now expects AI to affect even more areas than he previously thought. He recalls an early meeting with Nat Friedman and names Benchmark, his own firm and other investors he respects when asked about fund choices. These preferences and recollections reflect his personal experience and relationships.

He says one difficult decision was whether to match higher software valuations in 2016–2018. In retrospect, he believes Lead Edge was wrong to miss some of those deals. He also describes selling a small Shopify IPO allocation that would have been worth considerably more if held; the counterfactual reflects hindsight and omits the uncertainty investors faced at the time.

Green expects a major market downturn at some point within the next decade and hopes to have capital available to invest through it. He believes later generations of AI companies could benefit from both a correction and productivity improvements. The timing, severity and investment opportunities are forecasts, not known outcomes.