The massive capital expenditures required for the AI arms race are turning capital-light tech giants into capital-intensive operations. This shift will introduce significant depreciation and interest expenses onto their balance sheets, threatening to compress the exceptionally high profit margins that investors have come to expect.

Eclipse Ventures founder Lior Susan shares a quote from Sam Altman that flips a long-held venture assumption on its head. The massive compute and talent costs for foundational AI models mean that software—specifically AI—has become more capital-intensive than traditional hardware businesses, altering investment theses.

Building software traditionally required minimal capital. However, advanced AI development introduces high compute costs, with users reporting spending hundreds on a single project. This trend could re-erect financial barriers to entry in software, making it a capital-intensive endeavor similar to hardware.

The long-held belief from Fred Brooks' 'Mythical Man-Month'—that adding engineers slows projects—is now obsolete. With sufficient capital for GPUs and data, companies can compress years of software development into weeks, fundamentally changing competitive dynamics and making capital a primary weapon again.

Established metrics for evaluating software (high gross margins, capital-light) are obsolete in the AI paradigm. Top AI companies often exhibit opposite traits, like low margins due to inference costs, signaling the "death of spreadsheet investing."

The massive AI CapEx spending by hyperscalers is transforming the software industry's economics. The new model resembles capital-heavy industries like railroads or oil, moving away from the previous era's 80% margin software dream. Investors are now focused on the conversion cycle from spending to durable revenue.

Historically, software engineering required minimal capital—a laptop and internet. AI development now mirrors heavy industry, where the capital asset (like a $10M crane or $100M cargo ship) costs far more than the skilled operator. An engineer's compute budget can now dwarf their salary, changing team economics.

Software has long commanded premium valuations due to near-zero marginal distribution costs. AI breaks this model. The significant, variable cost of inference means expenses scale with usage, fundamentally altering software's economic profile and forcing valuations down toward those of traditional industries.

The software market is bifurcating. A few massive model companies (OpenAI, Anthropic) will be worth trillions and handle general tasks. The rest of the value will be in hyper-verticalized "for me" products. Mid-sized, general-purpose software companies will be squeezed out and struggle to compete.