The code does not lie, but it does hide. Last week’s earnings season stripped away the narrative fluff from big tech’s AI ambitions. The market didn’t reward the largest spenders—it punished them. Alphabet guided $205B in capex for 2025. Tesla burned cash at 142% of previous year. Intel and ServiceNow? They quietly won.
This isn’t about AI hype. It’s about capital efficiency. The market just executed a massive trust vote on who can turn silicon into profit without bleeding out.
Context
The Q1 earnings calls from four bellwethers revealed a stark divergence. ServiceNow’s subscription revenue grew 24.5%, cRPO hit $132B. Intel posted 25% revenue growth—fastest in 15 years—driven by Gaudi AI chips. Meanwhile, Alphabet’s cloud revenue jumped 82%, but free cash flow turned negative for the first time since 2004. Tesla beat on revenue but missed EPS badly, with capex surging 142%.
The market didn’t care about topline beats. It cared about the cost of those beats.
Core: Order Flow from Capex to Cash Flow
Let’s run the forensic analysis. I’ve audited smart contracts where a single overflow bug could drain a pool. Balance sheets are no different.
Alphabet – Cloud revenue grew $1.2B YoY. But capex guidance jumped to $205B. That’s a capex-to-cloud-incremental-revenue ratio of roughly 170x. Even if you factor in AI infrastructure for search and YouTube, the math doesn’t close. The market smelled it: stock dropped 8% post-earnings, option flows flipped bearish, analysts slashed targets.
Tesla – Delivery volume up 25% but ASP down. Automotive margin compressed. Free cash flow turned negative. The Dojo supercomputer and FSD are capital black holes. EPS missed by $0.33. Smart money rotated out: institutional outflows hit 3x the 30-day average.
ServiceNow – Capital-light AI integration. They don’t build TPUs. They use third-party models, fine-tune on customer data, and upcharge subscriptions. cRPO growth of 21% implies net revenue retention above 115%. The options market showed call accumulation despite a 3.7% post-earnings dip. Analysts raised targets.
Intel – Gaudi 3 is a direct beneficiary of NVIDIA’s supply constraints. Revenue per AI chip is lower than H100, but margin is positive. Capex is flat. The stock surged 15% on a volume breakout. Alpha hides in the friction of liquidity—Intel provides an alternative to CUDA lock-in.
Contrarian Angle: The Market Overpunished Alphabet
The crowd sees Alphabet’s $205B capex as reckless. I see it as a long-term bet with asymmetric upside—provided they maintain pricing power in cloud. Titanium, Google’s custom AI chip, has lower TCO per inference than NVIDIA. The problem is utilization. If Alphabet can fill those data centers with paying customers (including their own ad platform), the negative free cash flow is temporary.
But the market is right to demand proof. Volatility is the tax on uncertainty. Alphabet must show capex-to-revenue conversion within two quarters or risk becoming the next Cisco—spending billions on fiber that never paid off.
ServiceNow, on the other hand, has proven AI can be monetized without heavy capex. That’s the real signal. The model of renting AI inference from hyperscalers and layering it on sticky SaaS workflows generates high margins and predictable growth.
Takeaway
Check the gas, then check the truth. The next stage of the AI bull cycle won’t be won by those with the biggest infrastructure budgets. It will be won by those who can turn every dollar of capex into two dollars of revenue with 50% gross margin. Precision is the only hedge against chaos.
Watch ServiceNow’s cRPO growth and Intel’s AI chip revenue. Avoid companies where capex grows faster than free cash flow. The code does not lie—but the balance sheet does.
