In a wholesome innovation economic system, capital ought to usually align with society’s most pressing wants and the perfect risk-adjusted alternatives. Nevertheless, that’s not the case proper now. Enterprise capital has shifted dramatically towards synthetic intelligence and protection know-how, with deal sizes and valuations more and more resembling the speculative frenzy of the late Nineties dot-com bubble. On the similar time, vital points that may impression human well-being for the remainder of the century, local weather resilience, water shortage, and meals safety, stay chronically underfunded relative to their significance.
Meals and water costs are already flashing warning alerts: the FAO Meals Worth Index rose 6.9% year-over-year in 2025, whereas within the U.S., the all-food CPI climbed 23.6% between 2020 and 2024. These usually are not summary dangers; they’re eroding dwelling requirements at this time and reshaping world stability.
The AI frenzy: Development or bubble?
Synthetic intelligence has absorbed extra enterprise funding than another sector in historical past. In 2024, AI and machine studying startups attracted $131.5 billion globally, representing a 50% improve over the earlier 12 months. Within the U.S. alone, $109.1 billion in personal AI funding dwarfed China’s $9.3 billion, making this not only a market race however a geopolitical one.
Generative AI captured $33.9 billion of that complete in 2024, despite the fact that business income was simply $16 billion in the identical 12 months, with optimistic forecasts inserting it at $85 billion by 2029. That mismatch, greater than $100 billion invested yearly right into a market value a fraction of that at this time, is the traditional marker of bubble dynamics.
In the meantime, the underlying economics are strained. Coaching giant fashions can value tens of hundreds of thousands of {dollars} per run, with flagship Nvidia H100 GPUs priced at round $25,000 every. The infrastructure burden is equally daunting: McKinsey estimates world data-center capex will attain $6.7 trillion by 2030, with $5.2 trillion tied to AI workloads alone.
This isn’t an argument in opposition to AI itself. It’s an argument in opposition to capital piling into AI at valuations and burn charges that the actual economic system can not justify on enterprise timelines.
Protection: A increase with lengthy clocks
Protection know-how has additionally develop into a magnet for enterprise {dollars}. In 2024, the sector drew $9.1 billion throughout 228 offers, and in June 2025, Anduril raised $2.5 billion at a $30.5 billion valuation, one of many largest personal rounds on file.
The sector’s attraction is obvious: rising world tensions, trillion-dollar protection budgets, and governments signaling a willingness to work with startups. But the monetization hole is stark. A Wall Avenue Journal assessment discovered that the U.S. authorities invested solely $22 billion in contracts with the highest 100 national-security startups, lower than half of the $53 billion these corporations raised in enterprise capital.
Protection cycles are notoriously sluggish. Regulatory approvals, testing, and procurement stretch over years, usually a long time. For LPs anticipating fast exits, it is a mismatch between the persistence of the issue and the impatience of the capital.
The chance value
This isn’t only a ethical lapse; it’s a strategic blind spot. Water underpins meals, and meals methods underpin local weather stability. Agriculture alone accounts for 70% of worldwide freshwater withdrawals and about one-quarter of greenhouse gasoline emissions. If water methods falter, meals methods collapse, and with them, local weather objectives and nationwide safety alike.
The WHO and UN undertaking that by 2025, half the world’s inhabitants could possibly be dwelling in water-stressed areas, whereas the FAO reviews that 295 million folks already confronted acute meals insecurity in 2024, up almost 14 million from the 12 months earlier than. Droughts alone are draining $307 billion yearly from the worldwide economic system, and water shortage is projected to shave as much as 8% off world GDP by 2050. These are the sorts of systemic dangers that make AI GPU shortages look trivial.
However local weather tech already will get billions!
Local weather tech did certainly appeal to $56 billion within the 12 months to Q3 2024, in line with PwC. However that cash is unfold throughout power, mobility, business, buildings, and carbon administration. Inside that pie, water options accounted for simply 1-2% of climate-tech capital in 2024, and agri-food funding has been in steep decline since 2021.
When you settle for the World Sources Institute’s math, that feeding almost 10 billion folks sustainably would require a 50% improve in manufacturing alongside important emissions cuts, then at this time’s capital stack appears dangerously misaligned. Meals and water methods usually are not area of interest. They’re undervalued and undercapitalized, regardless of being the inspiration of survival.
A correction is inevitable. The FAO Meals Worth Index is up almost 7% 12 months over 12 months, utilities worldwide are imposing file double-digit water tariff hikes, and meals costs globally have risen 35% over the previous 5 years. Shortage is turning into the costliest commodity of the twenty first century.
The deep irony
Right here lies the irony. The very applied sciences absorbing enterprise capital rely immediately on the sources being uncared for.
Information facilities alone eat huge quantities of water for cooling. On common, a facility makes use of about 1.8 liters of water per kWh of power consumed, and a typical 1 MW information heart can draw 25.5 million liters yearly, sufficient to provide 300,000 folks for a day. In areas already dealing with water stress, the expansion of AI infrastructure is a direct competitor with households and agriculture.
And protection isn’t any completely different. Armies nonetheless march on their stomachs. Troops want meals to battle. If agriculture falters underneath water shortage and local weather stress, even trillion-dollar protection budgets can not paper over the scarcity of energy required to maintain militaries, not to mention civilians.
The engines of AI and protection finally relaxation on the soundness of meals and water methods. Ignoring that basis will not be solely shortsighted, it’s strategically self-defeating.
The market check
AI and protection are flush with capital, not as a result of their economics are confirmed however as a result of their narratives are irresistible. A flashy demo or a brand new protection contract can transfer billions in a single day. Meals and water improvements, in contrast, are incremental, technical, and slow-moving, however they ship tangible, compounding returns. Precision irrigation reduces prices 12 months after 12 months. Superior seeds improve resilience for many years. Water reuse and leak detection save billions in wasted sources.
These usually are not hype cycles; they’re the actual economic system.
What rebalancing ought to appear to be
If enterprise capital is severe about long-term returns and resilience, reallocation is important. Which means:
- Late-stage development fairness in precision irrigation with the identical depth we see in AI coding brokers.
- Billion-dollar debt services for municipal leak detection on par with data-center enlargement.
- Deep swimming pools of affected person capital for seed genetics, soil well being, and water reuse startups, not only for one other spherical of mannequin coaching runs.
In contrast to AI hype or protection cycles, demand for meals and water will not be elective, cyclical, or speculative. It’s as inevitable as biology.
The closing actuality verify
The purpose of enterprise capital is to not finance press releases; it’s to finance future money flows. Proper now, these money flows are imperiled much less by a scarcity of chatbots or protection drones and extra by the slow-motion disaster of droughts, floods, and soil degradation.
If LPs and GPs proceed mistaking spectacle for alternative, the correction shall be painful. A crowd of AI and protection unicorns will uncover their clients are slower than their burn, whereas undercapitalized meals and water methods face the local weather’s full brunt.
The arbitrage is hiding in plain sight: make investments the place demand is assured, externalities are optimistic, and shortage is pricing itself into the worldwide economic system. Enterprise capital exists to ship returns to LPs, to not chase mirages. The boom-or-bust bets on over-inflated bubbles in AI and protection are idiot’s gold; regular, grounded investments in water, meals, and local weather resilience will win the day.
