Hemant Taneja on Scaling General Catalyst, AI and the Future of Venture

General Catalyst CEO and managing director Hemant Taneja joins Harry Stebbings to discuss scaling an investment firm, AI adoption, labor and the future of venture capital. Taneja speaks as an investor in several of the companies discussed, including Stripe, Anthropic and Mistral.

Taneja sees his job as a dual role: he is both a partner investing in companies and CEO of General Catalyst as a business. He says venture remains the firm’s core and that its ambition is to be an excellent seed investor, because early relationships establish trust with founders.

Stebbings asks whether a large asset base makes early-stage work difficult to justify. Taneja says the firm emphasizes ownership and founder relationships rather than simply the size of the check, and has brought seed-focused teams into the organization to maintain that work.

The conventional way venture firms expand, Taneja argues, is by adding bigger funds, sectors and geographies. He thinks the more important innovation is changing what a firm offers founders so they can build larger, more durable businesses.

He believes scaling a traditional venture fund can hurt performance: having more money does not create more exceptional founders. His proposed response is to give founders additional tools and potentially help more companies become outliers; that is a strategy and aspiration, not a demonstrated causal result.

General Catalyst, he says, keeps venture funds at a size it believes can target four- to five-times returns, while offering separate creation and customer-value funds. Those other products support activities such as acquisitions and sales and marketing; the return multiple is a target rather than a guarantee.

Taneja says the firm’s best returns have come from seeding or creating companies, including Stripe and Livongo. He estimates Livongo returned roughly three to four times one fund and close to one times another, but explicitly says he would need to check the figures.

He says General Catalyst first invested in Stripe in 2010 and has invested 14 times over 15 years. The rationale for having a larger pool of capital, in his view, is to continue supporting businesses the firm believes can compound, even if it misses the seed round.

Asked whether he maps every important market and invests regardless of price, Taneja describes a balance between intention and serendipity. He recalls passing on Coinbase’s seed round because he did not understand the opportunity at the time, a reminder that future important categories are hard to identify.

Taneja identifies AI’s effect on jobs as an under-discussed shift. He thinks governments need to consider defense, healthcare, diffusion of AI into business and workforce reskilling together. His framework reflects General Catalyst’s own investment and advisory interests.

To transform an enterprise with AI, Taneja says, it needs usable data infrastructure, models adapted to its business, a plan for people working alongside agents, and leadership willing to drive change. He offers this as an explanation for why many pilots do not become operational transformations.

General Catalyst’s thesis is that some work previously offshored for labor savings could return closer to customers through AI productivity. Taneja says Crescendo, one of its companies, acquired a Philippine call center with 3,000 employees. He raises the question of what displaced workers would do; the future job impact is a prediction, not an observed outcome for all 3,000 people.

He estimates material labor effects could emerge over roughly five years as new companies assemble teams, gain customers and spread. He also recounts a consulting executive’s provocative hypothetical of an organization with 10,000 humans and 90,000 agents; he says such a change is not a forecast for the next five years.

Taneja worries that countries reliant on service exports could lose productivity gains to AI providers elsewhere, much as manufacturing shifted under globalization. His example of London professionals being replaced by agents is hypothetical. He argues that governments should plan for reskilling and local participation in the gains.

He praises conversations with officials in Singapore and Greece but says governments generally are not thinking enough about reskilling. Some leaders, he says, expect society to slow disruptive adoption; he doubts there is enough time to rely on that without deliberate action.

Taneja supports market-driven innovation but argues government has a role in preserving opportunity and the conditions under which capitalism works. He worries that technology’s gains could accrue to a narrow group instead of producing broad prosperity.

Stebbings points out that General Catalyst is an Anthropic investor and asks whether concentrated returns are inevitable. Taneja predicts relatively few foundation-model companies will reach large scale, but argues policy can support a wider ecosystem of businesses and services built on top.

Asked about the United States under the Trump administration, Taneja sees strengths in energy, AI, the market and entrepreneurship. He also worries that trade friction and changing international relationships could make it harder for American startups to become global leaders. This is his assessment, not an established forecast.

General Catalyst has backed defense companies in different regions, including Anduril and Helsing. Taneja says he expects regional leaders to compete globally and wonders whether new partnerships could combine their local advantages. His views reflect the firm’s investments; the cooperation he imagines is not an announced arrangement.

On US–China competition, he describes a bipolar technology environment and says adoption will depend partly on which models customers find useful. His view that the two countries are broadly comparable and separated by months is an assessment at the time of the interview.

Taneja sees a possible second-mover advantage in AI applications. A startup beginning on a newer model may build with stronger tools from day one, while an early company must decide whether to rebuild systems that already have customers. The advantage is theoretical and varies by execution.

General Catalyst invested in Anthropic at a $60 billion round, Taneja says, after seeing coding emerge as a business use case. He describes the round as a more concrete, risk-adjusted opportunity than a purely abstract bet on artificial general intelligence; the firm’s investment is central to his assessment.

He recalls Anthropic revenue at the time as under $1 billion, subject to his memory, and says subsequent growth exceeded the firm’s forecast. He says General Catalyst invested a few hundred million dollars at the $60 billion round and again at a later $180 billion valuation; these are his descriptions, not independently verified filings.

Taneja argues that the later round compared favorably on a revenue-multiple basis, while explicitly noting that the durability of model businesses is unclear. He says a future investment would depend on revenue, margins, profitability and scale rather than valuation alone.

He estimates very large potential markets in developer and white-collar labor and suggests Anthropic could be much larger if its trajectory continues. Those market sizes and potential multiples are estimates and conditional investor projections, not booked demand or promised returns.

Stebbings raises dilution and employee compensation in model companies. Taneja says he passed on an early OpenAI opportunity partly because he could not understand its structure, and discusses a reported allocation of value with an explicit caveat that he does not know whether the figures are correct.

Taneja views Microsoft’s OpenAI investment as strategically valuable for Azure, while suggesting the partners’ ambitions may create tension. Asked to compare Anthropic, he praises its focus and believes disciplined use of compute can matter as much as the amount raised. He is financially interested in Anthropic’s performance.

Stebbings asks what happens when extraordinary revenue growth slows. Taneja works through a hypothetical $9 billion to $27 billion revenue path and a 20-times multiple, implying a valuation above $500 billion. He explicitly says he is not predicting that outcome; it depends on the assumed revenue and multiple.

Taneja believes AI coding could command meaningful prices if it performs work comparable to that of an engineer, and says Anthropic has handled margins well. These are the investor’s claims and conditional economic argument, not independent evidence of durable margins or full job replacement.

He expects the number of major model providers to shrink and imagines a few global and regional providers. He also sees a very large potential market in labor spending. These are forecasts; existing labor expenditure is not automatically revenue available to AI companies.

Stebbings asks whether an AI slowdown could unsettle a market with concentrated technology valuations. Taneja thinks frontier progress could encounter technical obstacles, yet predicts that using existing capabilities for labor productivity will continue. His optimism is a scenario, not a guarantee about employment or the economy.

The conversation turns to Mistral, where a General Catalyst partner serves on the board. Taneja praises founder Arthur Mensch’s development as a CEO, says the company has improved its commercial approach, and expects demand for an open-source-oriented enterprise provider. This assessment carries the firm’s investment interest.

Asked whether national sovereignty can support a technology company, he points to defense suppliers as an analogy and argues that AI is strategically important. An analogy to defense does not establish that sovereignty alone will make a particular model company successful.

Revenue growth still matters, Taneja says, but the benchmarks he sees have changed. He discusses Mercor, whose seed round General Catalyst led, and cites rapid revenue growth; his enthusiasm and the numbers are company- and investor-reported claims, not proof that the growth will last.

Durability is the hard question. Taneja believes some rapidly growing AI companies will not endure, while expressing conviction in Mercor. He argues early market attention can help, but later entrants may have stronger technology and a more refined product proposition.

Taneja says the old triple-triple-double-double pattern is no longer sufficient for the fastest venture opportunities in his view. He distinguishes those from slower-growing businesses that may still be sound and worth supporting, including through General Catalyst’s customer-value fund.

Stebbings compares today’s uncertainty with pandemic-era categories whose growth proved difficult to forecast. Taneja calls this a period of peak ambiguity: rapid revenue growth can coexist with uncertain durability, while profitable structures do not necessarily reveal future growth.

His response is to use a long-term “true north.” For General Catalyst’s healthcare work, he says the aim is more proactive, affordable and accessible care; for European AI, he emphasizes resilience. These are stated goals, not independently verified evidence that any specific healthcare product improves outcomes or reduces costs.

He says investors and founders should decide what future they want to build while handling immediate pressures such as fear of missing out. General Catalyst frames its own offering around founder access to talent, policy expertise, distribution and capital; this is the firm’s proposition, not a measured comparison with peers.

On price, Taneja argues that an investor who believes a company can become exceptional should not dismiss it solely for seeming expensive. He says forecasts rarely land exactly as modeled and that the biggest returns come from outcomes better than expected. This is his investment philosophy, not a rule that price is irrelevant.

Stebbings challenges him with a hypothetical data business worth $2 billion to $4 billion at exit: an $80 million entry valuation produces different returns from $140 million. Taneja responds that markets can expand beyond initial estimates, citing early skepticism about Stripe, in which his firm holds a large position.

Taneja regrets occasions when General Catalyst did not invest more in companies that subsequently performed well. He says concentration and repeated investment in the best businesses can drive returns, but acknowledges that investors must be right about those businesses.

He estimates that perhaps 60–70% of the capital he has invested across more than 200 companies went into roughly ten companies. He says Stripe is his largest concentration at about $1 billion invested. These are approximate, self-reported figures, with concentration carrying obvious dependence on a small group of outcomes.

On investing across funds, he says a firm should consider each fund’s risk and give it meaningful exposure before bringing in another fund. He tries not to place more than 10–15% of a single fund in one company, while acknowledging that exceptional opportunities can lead to investment across multiple funds.

When a portfolio company goes public, Taneja considers whether continued involvement can help it compound and whether retaining shares serves the fund’s commitments. He says some limited partners sell distributed stock automatically, and large holdings also require attention to the pace of distribution.

For the strongest private companies, Taneja says secondary trading, employee liquidity, credit and acquisitions can resemble features of public markets. Other strong businesses may benefit from an IPO, while slower-growing firms can fall between the needs of public investors and venture funds. General Catalyst offers its customer-value fund to that group.

Stebbings asks whether companies staying private longer concentrates wealth by limiting ordinary investors’ access. Taneja calls access a serious issue and expects products to expand retail participation in private technology investments. He says General Catalyst intends to participate, giving it a commercial interest in that expansion.

Taneja says a venture firm should prioritize performance-based carry over management fees, and claims General Catalyst reinvests its fees in the business rather than distributing them. He frames compensation around alignment with performance; these are descriptions of his firm’s own practices.

General Catalyst’s offerings include venture, customer-value funding and company creation or rollups, he says. He is also studying infrastructure and energy because AI deployment needs power; he sees a potential path toward sustainability while noting near-term reliance on sources such as natural gas.

As the firm expanded beyond a single strategy, Taneja says it had to explain the approach to endowments and foundations that preferred specialized managers. He describes adding relationships with pensions and sovereign investors, including partnerships in which the firm advises regions as well as receiving capital.

Looking back, Taneja says he may have been too focused on choosing one winner in financial services and AI rather than investing across a whole trend. He has not yet adopted that indexing approach, but thinks it can make sense when the direction is clear and the individual winner is uncertain.

He regrets missing the early opportunity to invest in OpenAI, particularly because it would have given him a closer view of the technology. He says owning OpenAI would not necessarily have prevented an Anthropic investment; his comments are retrospective and reflect his existing Anthropic position.

Taneja describes a $500 million General Catalyst fund containing Livongo, Snap, Circle and Gusto. He estimates it could ultimately return 13 to 15 times capital, while acknowledging the story is not over; the figure is a projection, not a final realized return.

He imagines General Catalyst becoming a broader platform providing founders capital, distribution, policy access and wealth-management services. The firm has more than 300 people, he says, and he rejects a simple choice between a boutique investor and a large asset gatherer.

Taneja argues that assets under management can rise because an existing portfolio becomes valuable, not just because a firm raises more money. His preferred outcome is strong performance from relatively few exceptional businesses, rather than treating the amount raised as the sole measure of success.

He remembers meeting Stripe co-founder Patrick Collison in 2010 and asking about ideal customers. Collison replied that they had not yet been born, referring to the future developer movement. Taneja says the exchange exposed a gap in his own understanding and convinced him to back the founder.

Taneja tells Stebbings that losing competitive deals can be evidence an investor is meeting the best founders. He lost the Series A rounds of Stripe, Samsara and Snap before winning Gusto’s, and says persistence and later opportunities mattered more than dwelling on any single loss.

He says General Catalyst’s Stripe stake is below 10% and values its position at more than $5 billion, while describing about $1 billion invested earlier. His prediction that Stripe could become a trillion-dollar company in ten years is a forecast from a major investor, not a current valuation or promised return.

In the quick-fire round he says he has become more open to indexing across major trends. His leadership challenge is turning experience into something he can teach, and his advice to limited partners is to seek venture firms that innovate in how they support founders.

Taneja worries about aligning short-term business value with inclusive long-term prosperity. He says money is a byproduct of the impact he wants, encourages parents to help children ask questions, and thinks the value of particular educational paths may change as AI develops.

Looking ahead, he imagines General Catalyst investing hundreds of billions over the next two decades to help shape AI’s effect on society. That number is an aspiration, not committed capital. He hopes the transition ultimately leaves a positive legacy.