the ai bubble is a spending trap. here’s how to not get caught.
Hyperscalers are burning $700B a year on AI infrastructure with shrinking margins, while you might be losing thousands on unclaimed expenses. The lesson? Stop chasing scale. Start claiming what’s yours. ccLuca helps you do exactly that—no IT, no fuss, just your money back.
the numbers are staggering. $700 billion a year. that’s what hyperscalers like meta, microsoft, and alphabet are collectively spending on data centers and chips in 2026 alone. and for what? gary marcus, the nyu professor who’s been calling the ai hype since day one, puts it bluntly: this looks a lot like the airline industry. thin margins, brutal competition, and a race to the bottom.
"This has led to price wars coexisting with high operating expenses... and low or even negative margins, since all are building more or less the same product."
he’s not wrong. when everyone builds the same thing, the only differentiator becomes price. and price wars destroy value. fast.
the $700 billion question
marcus’s argument hits a nerve because it’s not just about big tech. it’s about a mindset. the idea that throwing more money at a problem—more chips, more data, more servers—automatically solves it. but it doesn’t. not when your product still hallucinates facts. not when open-source models from china are catching up. not when your own customers start looking for cheaper alternatives.
microsoft, for example, is already testing deepseek—a chinese open-source model—for its copilot coworker. why? because costs are ballooning. charles lamanna, microsoft’s evp for copilot, admitted:
"We have users who do hundreds of tasks a week, which is great—they're way productive—but the consequence is the costs can go very high."
so the hyperscalers are stuck. they can’t stop spending because they’re afraid of falling behind. but they can’t keep spending because margins are evaporating. it’s a trap.
the same trap, just smaller
you don’t need to be a hyperscaler to fall into this trap. it happens at every level. freelancers, small teams, even individuals. we all have expenses we forget to claim. receipts that pile up. small amounts that, over a year, add up to something real.
think about it: the average person leaves hundreds—sometimes thousands—of euros on the table every year. unclaimed business expenses, tax-deductible purchases, reimbursable work costs. it’s not $700 billion, but it’s your money. and you’re leaving it there.
the minimal solution
here’s where the berlin tech minimalist in me gets excited. you don’t need a massive system to fix this. you don’t need enterprise software, a dedicated it team, or a complicated process. you just need a tool that works.
that’s why ccLuca exists. snap a photo, get ai-extracted data in 3 seconds, generate expense reports instantly. no setup. no training. no bullshit. it’s built for individuals and small teams who want to stop losing money to administrative friction.
The expenses you forget to claim could buy you an iPhone every year.
that’s not a marketing line. it’s a fact. and unlike the hyperscalers, you don’t need to spend billions to fix it. you just need to start.
what we can learn from the ai bubble
the hyperscaler story is a cautionary tale about scale without strategy. about spending without thinking. about building the same thing as everyone else and hoping for a different result.
don’t be like them. be efficient. be intentional. claim what’s yours.
source: hyperscalers could end up resembling airlines—plagued by small margins, intense ...