Tobi Lütke on Building Shopify, AI and the Future of Work

Shopify co-founder and CEO Tobi Lütke joins Harry Stebbings for a wide-ranging conversation about building companies, AI and public policy. Lütke offers personal views and predictions, including contested claims about layoffs, charity, government and climate policy; these are not established findings of the interview.

Stebbings asks whether entrepreneurs are driven more by fear of losing or hunger to win. Lütke says he is still working out his answer, but sees a short-term focus in both orientations. He prefers a longer perspective: helping colleagues grow or giving them difficult tasks can compound over years, changing the way a leader approaches decisions.

Lütke uses the Enneagram personality framework to explain his own style, describing himself as an “eight” who wants to fix what is wrong. He believes companies often screen out people who challenge established practices, while founders can create space for that temperament. This is his interpretation of a personality framework and business experience, not a validated rule for identifying leaders.

Asked about the loneliness of leadership, Lütke says company builders are often unusual, high-variance people and that polished public images conceal the difficult parts of the job. He enjoys much about his position but says he did not originally want to be CEO. He frames his job partly as creating room for others to do the work he wishes he could do himself.

He says leading a product company requires control over decisions whose short-term business costs may precede longer-term benefits. A genuine long-term position can mean accepting weaker numbers for a period, not merely talking about future results. Those choices, he says, can be uncomfortable.

Stebbings asks whether a public company has the freedom to think long term. Lütke distinguishes trusted and untrusted companies: in his view, earning public investors’ trust gives a company room to pursue a long horizon. He says Shopify went public relatively early and benefited from giving investors time to see its execution.

Discussing the IPO, Lütke recalls that underwriting incentives and the allocation process mattered when setting a price. He studied how bankers were paid and told them that his responsibility was to Shopify. His account is a founder’s recollection of the process, including financial figures he says he may not remember exactly.

Lütke describes the tense meeting to set the offering price and says he raised it above the original range because of demand. The first trade was higher still, in his recollection, showing that the market had a different assessment. He believes early investors who stayed with Shopify benefited from its later growth.

Asked about staffing five years from now, Lütke says he hopes Shopify might remain around its then-estimated 7,500–8,000 people while becoming much more productive. The suggested 100-fold productivity is an aspiration, not a forecast backed by a method in the conversation. He expects AI to make starting a business easier and describes entrepreneurship as potentially benefiting from the technology.

Stebbings asks whether current layoffs reflect AI efficiency or pandemic-era overhiring. Lütke attributes the layoffs he is discussing to overhiring and argues that AI is becoming a convenient scapegoat, while acknowledging that AI will change the mix of jobs. He does not provide company-by-company evidence, so his explanation should not be treated as establishing the cause of layoffs generally.

Stebbings challenges his optimism, pointing out that agents can perform many execution-oriented tasks. Lütke argues that work consisting solely of tasks assigned by others is not a good job and values work with agency and the ability to create. This is his view of job quality; it does not resolve how displaced workers would find alternative jobs or income.

Lütke predicts that cheaper products, higher purchasing power and new ways to exchange value will give people more options. He offers the example of a programmer who built the Ghostty terminal after selling a company, suggesting that people will find new kinds of productive work. These are possibilities he anticipates, not demonstrated outcomes for the labor market.

As another example of jobs created around an activity, he points to Formula 1 and the engineers who work on engines, chassis and aerodynamics. Stebbings adds that storytelling helped broaden interest in the sport. Lütke thinks combining existing elements in new ways can create valuable activities and employment.

Lütke argues that expensive software has limited what people could build and that new businesses turn potential resources into products others can use. He presents entrepreneurship and markets as mechanisms for creating and distributing wealth. Stebbings then asks whether hostility toward wealth has increased in the United States.

Lütke says wealthy people and institutions deserve scrutiny, but argues that public debate often misidentifies who contributes value. He praises Elon Musk’s products, including Tesla vehicles, Starlink and X, as examples of services he personally finds useful. His praise is a judgment about those products and their creator, not an assessment of every related controversy.

He draws a distinction between wealth acquired by building a business and wealth inherited or received through divorce, arguing that those managing inherited wealth should be scrutinized for its use. He also asserts that media can distort public judgments. Stebbings challenges him with the example of a large charitable gift and asks whether criticism could discourage giving.

Lütke responds that spending is a vote for a product and its supply chain, and argues that allocating money should be judged by its second- and third-order effects. In his account, purchases signal demand and allow businesses to expand. Stebbings and Lütke disagree about whether scrutiny of charitable donations might deter useful giving.

Lütke says donations should be evaluated by their actual effects rather than their stated intentions. Stebbings argues that donors may lack time for due diligence and could simply stop giving if every gift attracts heavy scrutiny. Lütke maintains that effective organizations will attract support, though the exchange does not establish whether that is generally true.

Lütke expresses skepticism of nonprofit organizations because, in his view, they lack the market feedback mechanism available to for-profit enterprises. He also acknowledges that good charities exist. His broad argument is a political and economic position, not proof that nonprofits cannot measure outcomes or that markets always do so accurately.

Asked about government intervention, he cites Friedrich List and argues that government should establish rules that encourage beneficial competition. He also assigns it roles in security, property rights and infrastructure. He sharply criticizes government efficiency, including a numerical claim about costs that he does not substantiate in the interview.

Stebbings asks what could help European growth. Lütke first discusses Canada, arguing that its resources could support more domestic development and processing. He believes the country could build more mines and infrastructure, a policy preference that does not address the full environmental, Indigenous-rights or economic trade-offs in the interview.

Stebbings asks about Canadian Prime Minister Mark Carney’s stance toward Donald Trump. Lütke praises a speech but disagrees with its approach to the United States, and uses a disparaging label for what he sees as an excessive Canadian reaction to Trump. His characterization and his approximate polling figure are his assertions, not independently established here.

Lütke argues that Canada can diversify its relationships with Europe and Asia while retaining close economic ties to the United States. He calls for pipelines, industry and more processing of domestic resources. Stebbings raises concerns about Chinese AI models and their use by US startups.

Lütke says he worries that restrictions on children’s access to AI could push them toward models with a single political perspective. He asserts that Chinese models can reflect a collectivist worldview and gives an example involving Tiananmen Square. These are his concerns and generalizations; the interview does not test particular models or establish what children would use.

Asked about social media at home, Lütke says his children are uninterested in having accounts or phones and instead use computers and play games. He says their family experience is unusual and offers little general guidance. He also cautions that experiences may differ among children.

On Europe, Lütke criticizes environmental restrictions and opposition to nuclear power, arguing that they can make infrastructure difficult to build. He favors government rules that support internal markets and competition, while acknowledging that governments provide services and fund infrastructure. His criticism is a policy view, not evidence that particular environmental safeguards are unnecessary.

He argues that infrastructure can create value over periods longer than most businesses can plan for, making it an important government role. He sees economic growth as the foundation for later decisions about how to use shared resources. His claims about returns from airports and other projects are illustrative estimates, not audited figures.

Asked about his costliest mistake, Lütke points to Shopify’s move into physical logistics and warehousing before AI improved rapidly. The company could not pursue both priorities in the same way and abandoned that direction. He says he once defended the original decision, but now thinks information available at the time might have led him to choose differently; the consequences affected employees.

Lütke prefers to describe a leader’s role as adding “heat” rather than injecting chaos: energy can help people make something new. He says he concentrates on areas that lack that energy, leaving teams already moving quickly more room. The metaphor reflects his own leadership style, not a universal prescription.

A CEO who concentrates on problems can spend weeks hearing only difficult news, he says. Lütke keeps a collection of encouraging comments from people he respects to revisit during those periods. He describes himself as deeply emotionally attached to Shopify, even though he does not follow its share price closely.

Lütke distinguishes the stock ticker, which he sees as the market’s estimate of company value, from the work of building that value. He remembers a large share-price rise early in COVID and says it did not mean the company had suddenly become proportionately smarter. He notes that other industries may place greater emphasis on investor relations.

Asked what he has changed his mind about, Lütke says he initially expected new engineers without established habits to have a special advantage in AI-native work. He has since seen the value of senior engineers’ experience in steering AI toward the right solution. He still describes some younger engineers as exceptional and says he does not know all the implications.

Lütke predicts that a role combining product, design and engineering may emerge around context engineering and coordinating people and AI. He compares clear instructions to good communication and says managers can bring useful experience to directing AI systems. He emphasizes that the eventual role may look different from his description.

He describes River, an AI system used through Slack that can work with Shopify’s large code repository while people steer it. He says the system chose its name after being asked. Lütke reports that well over half of Shopify’s code is now AI-generated and that some top engineers have not personally written code this year; he does not give the measurement method.

Lütke says he still enjoys writing code and sometimes works directly on data structures and persistence before steering AI through other parts. He believes his engineering background helps him ask precise questions and evaluate system design. This is his account of his workflow, rather than a claim that AI output needs no review.

On university, Lütke says he would encourage a child to pursue a demanding degree when it puts them among people deeply interested in the same subject. He also thinks joining a small company could be better if the person has skills the company values. Both paths, in his view, depend on the opportunity and the individual.

Stebbings asks about nepotism. Lütke works through the question, ultimately saying he sees no redeeming value in it and regards double-blind assessment of merit as an ideal. He criticizes some Canadian academic hiring practices while conceding that efforts to overcome barriers can sometimes identify excellent candidates; his broad characterization is not independently established in the interview.

His closing advice is that people can try to do things rather than assume an existing system rules them out. He adds a condition: experiments should not impose harm on other people, and he urges people to look for positive-sum opportunities. Acting, he says, also produces information about what works and why a system may discourage an idea.

After the apparent wrap-up, Lütke says he hopes people who disagree will continue listening and that conversation should leave room for people to be wrong. He argues for institutions that allow people to build while holding them accountable for their actions. He also discloses owning about 6% of Shopify and argues that customers and other shareholders benefit from its growth.

He contends that purchasing products is a way people influence the economy and defends company builders against broad accusations that wealth necessarily comes from taking from others. He welcomes substantive criticism but objects to what he calls lazy or bad-faith commentary. These are his arguments about wealth and criticism, not findings that settle the distributional effects of any particular business.

Lütke criticizes what he sees as alarmist public claims about climate and calls for corrections when predictions are wrong. He says climate change exists, while disputing unnamed past predictions about ice. His comments do not evaluate the evidence on present climate change or its risks, and the interview does not identify the specific predictions he invokes.

Stebbings asks about people turning to AI tools to check claims. Lütke is optimistic that AI could reduce the imbalance between making an allegation and investigating it. Stebbings points out that malicious actors can also use AI; Lütke agrees, mentioning voice cloning and software security as risks while predicting that AI could also help defenders.

Lütke proposes comparing responses from models trained in different countries as one possible way to examine claims, and imagines building such a tool quickly. He expects better tools to help people challenge bad-faith information. That is a proposal and prediction, not evidence that a council of models would reliably verify facts.