The hosts challenge the conventional accounting of AI training runs as R&D (OpEx). They propose viewing a trained model as a capital asset (CapEx) with a multi-year lifespan, capable of generating revenue like a profitable mini-company. This re-framing is critical for valuation, as a company could have a long tail of profitable legacy models serving niche user bases.

Aiming for complete feature parity between an old and new system is a trap. It forces the business to halt innovation for an extended period, and by the time the 'perfect' replacement is ready, the market has moved on, rendering the new system already outdated.

Citing Salesforce veteran George Hu, Halligan notes that in hypergrowth, nothing scales for long. Any new system, process, or even role has a three-year lifespan before it breaks and needs to be replaced. This mindset normalizes constant change and helps leaders anticipate inevitable breaking points.

Engineers may advocate for modernizing a functional legacy system not for business needs, but to add popular new frameworks to their resumes. This 'RDD' leads to wasted budget on projects that don't deliver real customer value, a phenomenon labeled Resume-Driven Development.

Hyperscalers are extending depreciation schedules for AI hardware. While this may look like "cooking the books" to inflate earnings, it's justified by the reality that even 7-8 year old TPUs and GPUs are still running at 100% utilization for less complex AI tasks, making them valuable for longer and validating the accounting change.

The debate over AI chip depreciation highlights a flaw in traditional accounting. GAAP was designed for physical assets with predictable lifecycles, not for digital infrastructure like GPUs whose value creation is dynamic. This mismatch leads to accusations of financial manipulation where firms are simply following outdated rules.

The useful life of an AI chip isn't a fixed period. It ends only when a new generation offers such a significant performance and efficiency boost that it becomes more economical to replace fully paid-off, older hardware. Slower generational improvements mean longer depreciation cycles.

Some tech companies have doubled the depreciable life of their AI hardware (e.g., from 3 to 6 years) for accounting purposes. This inflates reported earnings, but it contradicts the economic reality that rapid innovation is shortening the chips' actual useful life, creating a significant red flag for earnings quality.

To successfully advocate for a working legacy system against modernization pressure, you must be deeply aligned with its profit and loss. If someone else controls the P&L, your customer-centric arguments will be overruled by financial or political motivations, making your position untenable.