When firms go low, employees go solo

PHILADELPHIA – On the day Meta began implementing its plan to lay off a tenth of its workforce, a one-time employee, Xiaoyin Qu, saw her chance.

“We rented an LED truck and drove inside Meta’s headquarters,” she said.

The truck’s screens flashed a blunt message at departing employees of the social media giant. “Fired? Start a company before lunch.”

The stunt appears to have worked. Qu says dozens of Meta employees reached out to her start-up, which provides an AI operating system for one-person firms. 

The exact number is hard to verify, but the upshot appears to be that layoffs can perhaps be reframed as a beginning.

China-born Qu, the 33-year-old founder of HeyBoss.AI, represents a new kind of entrepreneur emerging across Silicon Valley even as heavy tech layoffs dominate the news. Her company, based in Redwood City, California, helps users build businesses using teams of AI “executives”. In 2025, she made a splash by stepping aside in favour of an AI CEO she named Astra. 

Astra, which helped Qu with negotiations to secure a US$3.5 million (S$4.5 million) seed round backed by the OpenAI Startup Fund in 2025, also runs her other ventures.

Qu’s latest product, Tycoon.us, pushes the idea of an AI chief executive further. An aspiring founder can prompt Astra with a goal – “10x traffic”, for instance – and it generates a plan, assigns AI agents, tracks execution and escalates decisions to the founder only when needed. 

The appeal is obvious in a moment when workers increasingly feel exposed and many see themselves as dispensable cost centres that can be cut to boost corporate margins. The result is a strange loop; AI is helping solve a problem that AI itself helped create.

The scale of AI-driven layoffs is disputed but there is a sense of where things might be headed.

Over 100 tech companies cut more than 115,000 jobs in the first quarter of 2026, up sharply from a year earlier and the highest quarterly total since early 2023, according to tracking site Layoffs.fyi. 

Meta alone eliminated roughly 8,000 roles, including in Singapore. Microsoft has offered buyouts to thousands. Amazon and Oracle have made similar moves. Google is rumoured to follow.

At the same time, these companies are pouring hundreds of billions of dollars into AI infrastructure, building data centres, chips and models. The juxtaposition is stark: shrinking payrolls alongside record capital expenditure.

There is not a single authoritative total for US “jobs lost to AI” for 2026, but the figure is in the tens of thousands, not millions. Still, AI-led cuts are rising.

The data fills in some gaps. AI was the most cited reason for 83,387 job cuts in April by US-based employers, according to outplacement and executive coaching firm Challenger, Gray & Christmas, which produces widely cited reports on layoffs. One in four of those cuts, or 21,490 positions, was lost to AI in April, said the May 7 report. 

In the first four months of 2026, AI was associated with nearly 50,000 job cuts, roughly 17 per cent of the total layoffs and the third-leading cause for retrenchments.

In contrast, US companies in 2025 announced 1.2 million job cuts. AI was mentioned as a reason in 55,000, or just 4.5 per cent.

Also telling is where the cuts are concentrated. The Bureau of Labor Statistics reported in May that employment in the “information” sector, which includes tech, has fallen about 11 per cent from its 2022 peak. Outside that sector, the impact is far less visible.

A debate about causality is also growing. Are workers being replaced by AI or are companies cutting jobs to fund AI?

In many cases, the latter may be closer to the truth. Building and running AI systems is expensive. Firms are reallocating resources, shifting from labour to capital. 

Oracle offers a stark example: strong revenue growth, surging AI business lines and 30,000 layoffs alongside a massive US$50 billion commitment to data centres.

Some analysts have noted “AI-washing” – companies exaggerating their AI adoption to appear more innovative and using it as justification for decisions driven by more traditional pressures, like overhiring. Economists are convinced, however, that the US labour market remains resilient. 

Monthly layoffs in the US routinely exceed 1.5 million across all sectors, making AI-related cuts a small fraction of the total, pointed out Guy Berger, a macroeconomist and senior fellow at the Philadelphia-based Burning Glass Institute, a nonprofit focused on studying the future of work. 

But the psychological impact is outsized. In Silicon Valley especially, the mood has darkened. Even highly paid workers increasingly feel that their roles are temporary, their skills expendable.

That anxiety is reinforced by leadership messaging. When CEOs speak of efficiency, automation and AI-native organisations, employees hear a more threatening message: fewer of them will be needed.

Out of that uncertainty, a new model is taking shape: the one-person company.

Qu is one example, but she is far from alone. Across fintech, media, software and e-commerce, small teams – often just one or two people – are building products that would have required dozens of employees only a few years ago.

The enabling force is not just AI capability – its raw intelligence, reasoning, coding, et cetera – but also AI orchestration, which is how that capability is used to get work done. Tools now handle coding, marketing, customer support, analytics, and even decision-making. What remains for the human founder is defining vision: choosing what to build and why.

“A useful analogy is the shift that followed the arrival of cloud computing and app stores in the 2000s,” said David Yin, a Singaporean who is a partner at Informed Ventures. The Menlo Park, California-based venture capital fund has poured around US$300 million into early healthcare and fintech bets. 

Those technologies dramatically lowered the cost of starting a company by removing infrastructure and distribution barriers, Yin said. AI is now doing the same for production itself. The consequences are twofold, he said. 

First, the number of companies is likely to explode. When the marginal cost of building drops towards zero, experimentation rises. More people can try, fail and try again. Second, the average company size shrinks. Tasks that once required teams are now handled by systems. A founder, for example, can operate with a handful of contractors and a network of AI agents instead of full-time employees.

But the benefits come with limits; lower barriers to entry also mean more competition. Building a product is easier, but building a business remains the challenge it has always been. Many AI-generated products struggle to gain attention or revenue.

Lihong Wang’s months-old New York-based startup, Freeport Markets, illustrates this new model of entrepreneurship, with its promise and constraints.

Less than a year after leaving a major trading firm in 2025, the Guangzhou-born 24-year-old launched a platform aimed at bridging the gap between retail and professional investors. Freeport combines AI-driven market analysis with a platform that allows users to trade derivatives tied to stock prices around the clock, even outside market hours. An added attraction is a newsfeed, created by AI analysts, that updates users with market-moving news and actionable trade ideas. 

“For example, with the current Iran situation, a lot of news happens on the weekends, when people are not able to trade off that news via traditional brokerages, whereas they can trade 24/7 with us,” Wang said. His platform also allows users to trade tokenised shares of companies that have yet to make a stock market debut, like SpaceX, OpenAI and Anthropic.

The company reached over 10,000 downloads and tens of millions in trading volume within months of its launch. Its core team remains minimal: Wang, his co-founder and a full-time engineer.

AI handles much of the rest, but not everything. “What AI is not able to do right now is create independent systems at scale. It still needs some prompting, for example, to know how to set up servers in the right architecture that allows scaling, so the app doesn’t break when you have 10,000 concurrent users. Having a really good backend engineer who does that for us is very value-additive,” he said.

While Wang is the CEO of his firm, even that function can be almost outsourced. Qu says her AI CEO Astra, conceptualised as a female, can manage hundreds of companies while a human would struggle with more than a few. 

“But not just that, Astra can learn from experience,” Qu said. “For example, when we try different marketing campaigns, she will document the decisions, see what works, what doesn’t. And she can learn across different companies, without compromising privacy, and get better and better.

“From an execution standpoint and even from a leadership standpoint, I think AI will do a better job than humans as it learns more. So then our value as humans is to generate the vision, like Steve Jobs did for Apple or Elon Musk for his companies.”

Anxiety about layoffs, one’s relevance and the pace of change is real.

“The pitch has been that AI is going to make us a lot more productive,” said Berger. “Therefore, we’re going to need fewer workers for the same output. To pessimists, that means layoffs. Optimists are going to say technology has done this for a long time, and it just means that people end up doing different kinds of jobs,” he said.

“That’s conditional on improvements in productivity that I don’t think we’re seeing yet. If the returns to AI don’t materialise, companies might end up having to hire back their workers.”

For most workers, the immediate question is adaptation.

As companies become more explicit about using AI for tasks like customer support, scheduling, and data processing, the challenge is not just displacement but transition. How quickly can workers move into roles that complement AI rather than compete with it? Should there be stricter oversight to ensure they don’t get the short end of the bargain?

Some policy responses are beginning to take shape. 

Proposed legislation in the US Congress would require greater transparency on AI-driven layoffs. One Bill calls for the Department of Labor to collect and publish quarterly data on layoffs, hiring and retraining tied to AI adoption. 

Trade unions are pushing for a more proactive role, wanting a seat at the table rather than reacting after layoffs. AI deployment should be negotiated rather than imposed, they argue.

There is also the intriguing idea of wealth redistribution to cope with the feared AI-driven mass layoffs. In April, OpenAI’s Sam Altman proposed a “public wealth fund”, financed by tech companies, which will give all citizens a stake in AI-driven economic growth. 

Elon Musk, who has spoken of putting data centres in space and whose company makes humanoid robots, has advocated a “universal high income” or cash infusions from the federal government to citizens as offsets.

Berger, an expert on US labour markets, said it would not be a bad thing to shore up the safety net for Americans. “If this technology is as impactful as its proponents think, it would be a shame not to harvest it for the gains, including redistribution, instead of slowing its adoption,” he said.

“I don’t buy that there’s a world where people are losing tonnes of jobs, this technology is happening, and we don’t have the resources to solve these problems. But the question is: Does it take a crisis to get there, or does it happen organically and smoothly?” 

For the tens of thousands impacted by tech layoffs, the crisis is now. Yin’s guess is that some will gravitate towards opportunities in AI-native firms. For others, the boundary between employee and entrepreneur is dissolving as tools that once required organisations are increasingly available to individuals.

AI is producing an unsettling change that can feel like a loss for most and like leverage to a few. 

“It’s unfortunate that people lose jobs because of the disruption from AI,” Wang said. “But if AI did not exist, my company probably would not exist because it would be so difficult to build. It would have taken anywhere between three and five times as long for development if we did not have AI. And we’d probably have needed two to three more engineers to help us.

“So, the three jobs that we have in the company right now were created because of AI.”