$242 billion bought us nothing

In the first quarter of 2026, investors poured roughly $297 billion into startups worldwide. Of that, $242 billion, about 81%, went to AI companies. That is not a trend. That is a market rewriting itself in real time. Four deals accounted for most of the damage. OpenAI closed $122 billion at an $852 billion valuation. Anthropic raised $30 billion. xAI pulled in $20 billion. Waymo added $16 billion. Together, those four rounds totalled $188 billion, or about 63% of all global venture investment in a single quarter. The numbers are so large they stop feeling like numbers. They feel like religion. And the question worth asking is not whether AI is real, it obviously is, but whether capital deployed at this scale is actually producing anything.

The production gap

According to Deloitte's 2026 State of AI in the Enterprise report, 75% of companies say they plan to invest in agentic AI. But only 11% have AI agents running in production. That gap, between intention and execution, is where billions of dollars are quietly vanishing. This is not a small discrepancy. It means that for every company shipping an AI agent to real users, roughly eight more are stuck somewhere between proof-of-concept and PowerPoint. The money is moving, but the output is not. Gartner's projections tell a similar story from a different angle. They estimate that 40% of enterprise applications will include task-specific AI agents by the end of 2026, up from less than 5% in 2025. That is an ambitious forecast, but even if it lands, it means the majority of enterprise software will still have no AI agent integration by year's end, despite a quarter-trillion dollars flooding the space.

We have seen this before

The pattern is familiar. Capital floods into a category before unit economics exist, before most companies have figured out how to turn the technology into a product that someone will pay for on a recurring basis. In 1999, U.S. venture capital investments totalled $54 billion, with 62% going to unprofitable companies. NASDAQ saw 442 IPOs in 2000 alone. Then the index lost 78% of its value over the next two and a half years. The fibre optic cables that were overbuilt during the dot-com frenzy eventually became the backbone of the modern internet, but not before wiping out the companies that built them. Crypto followed the same arc. In 2021, crypto venture capital hit $33.8 billion, nearly 5% of all VC that year. Money poured into projects with no revenue model, token launches with no users, and protocols that solved problems nobody had. By 2023 and 2024, funding had collapsed. The companies that survived were the ones with actual products and actual customers. The parallel is not that AI is fake. The internet was not fake in 1999. Blockchain was not fake in 2021. The parallel is that capital is a terrible proxy for progress.

Buying compute, not customers

Most of the $242 billion is not going toward acquiring customers or building products. It is going toward infrastructure, compute, and the raw materials of model training. Morgan Stanley Research estimates roughly $2.9 trillion in global data centre construction costs through 2028. ARK Invest projects nearly $1.5 trillion in annual AI infrastructure spending by 2030. These are staggering numbers, and the firms making these projections argue this is industrial buildout, not speculation. Maybe. The counter-argument is that cloud computing followed a similar trajectory. Amazon, Google, and Microsoft spent years building infrastructure that seemed excessive relative to demand. Then demand caught up, and the infrastructure became the product. AWS alone now generates over $100 billion in annual revenue. But there is a crucial difference. The cloud buildout was led by three or four companies with existing cash flows, existing customers, and existing distribution. The AI buildout is being funded largely by venture capital, which means the money comes with expectations of returns on a venture timeline. When OpenAI raises $122 billion at an $852 billion valuation, the implicit promise is not "we will be a good utility in twenty years." It is "we will justify this valuation within a few years, probably via IPO." That is a different bet entirely.

The oxygen problem

When four companies raise $188 billion in a single quarter, the gravitational pull distorts everything around it. For indie builders and smaller startups, the effect is suffocating. Talent gets priced out. Compute costs, while falling, still favour companies that can negotiate billion-dollar cloud contracts. Media attention narrows to the frontier labs. Investors start pattern-matching on "who is the next OpenAI" instead of asking "what problem does this solve." Early-stage funding is technically up. Crunchbase data shows $41.3 billion across 1,800 early-stage deals in Q1 2026, with seed funding reaching $12 billion (up 31% year-over-year). But the number of actual seed deals dropped by 30%. Investors are writing bigger cheques to fewer companies. The bar for a first round has gone up, not because the technology is harder, but because the narrative has shifted toward scale. This is the part of the cycle where good ideas die quietly. Not because they lack merit, but because they lack the capital to compete with companies that have raised more money than most countries spend on education.

The Singapore question

For anyone building from a smaller ecosystem, the concentration of capital is not just an abstract concern. It is an existential one. Singapore punches well above its weight. It ranked 4th globally and 1st in Asia in the 2025 Global Startup Ecosystem Index. It captured 91% of all Southeast Asian venture capital, with funding jumping 202% year-over-year. The government has committed S$1 billion to the Startup SG Equity scheme, launched a S$1.5 billion Anchor Fund, and made AI a national priority through the National AI Strategy 2.0. But context matters. Singapore's entire monthly VC activity fluctuates between $200 million and $400 million in a typical month. OpenAI raised that much in the time it takes to read this sentence. The total Southeast Asian venture ecosystem closed 2025 with $5.4 billion raised. That is less than 2% of what AI companies alone raised in Q1 2026. The question for small ecosystems is not whether they can compete on capital. They cannot. The question is whether they can compete on something else entirely: speed of execution, regulatory clarity, proximity to underserved markets, or simply better taste in what to build. Google is expanding AI investments in Singapore. AI startups are increasingly incorporating there for its IP protection, regulatory framework, and access to Southeast Asia's 700-million-person market. These are real advantages. But they are advantages that compound slowly, while capital compounds fast.

Capital is not progress

The point is not that AI is a bubble. The technology is transformative, the use cases are real, and the productivity gains for individuals and teams are already measurable. The point is that $242 billion in a single quarter does not mean we are $242 billion closer to a future where AI makes most people's lives materially better. Most of that money is buying GPUs, not building products. Most of it is funding valuation growth, not customer growth. Most of it is concentrated in four companies that are racing each other to build the biggest model, not the most useful one. The companies that will matter in five years are probably not the ones raising the most money today. They are the ones shipping products that work, solving problems that exist, and building businesses that generate more revenue than they consume in compute. History does not punish investment in real technology. It punishes the belief that spending and progress are the same thing.