AI Adviser Zack Kass Says Software May Have Been in a Bubble

"The market certainly is excited and continues to be excited about what Jensen is saying and what NVIDIA is doing for good reason... we're still in the beginning of realizing how much value we could create... with unmetered intelligence." Despite fears of a bubble, the fundamental demand for the "utility" of intelligence is just starting. NVIDIA provides the "power plant" infrastructure for this new utility. The trend is secular, not cyclical. LONG. The infrastructure build-out phase is not over. Hyperscaler capex cuts or geopolitical restrictions on chip exports.

"Maybe software was the bubble all along... monolithic pieces of software... We're seeing plumbing new depths for CRM for Salesforce... Snowflake did turn lower." Traditional SaaS companies trade at high multiples (30x) based on seat licenses and user interfaces. Agentic AI ("unmetered intelligence") reduces the need for humans to operate complex software, thereby destroying the pricing power and moat of legacy "System of Record" companies. SHORT. The market is repricing these assets as their utility is cannibalized by AI agents. AI integration into these platforms could be successful enough to defend their moats, or oversold conditions could trigger a technical bounce.

"If value comes out of... software in particular, it's gonna go somewhere else... My bet for the last four years has been bio and life sciences... molecular particle material, construction, and anything building in the real world, hospitality, entertainment." This is Second-Order Thinking. As the cost of intelligence (software/analysis) drops to near zero, the sectors that were previously limited by high R&D or coordination costs (Biotech, Construction, Materials) become the primary beneficiaries. The value lost by SaaS flows into "Atoms" (physical world) rather than "Bits." LONG. Rotate exposure from pure software into physical industries and novel sciences. Regulatory hurdles in bio/construction or a general recession slowing down physical economy spending.

Interviewer notes trucking/logistics companies have "really benefited from AI and machine learning." Kass agrees, noting AI creates value in the real world. Logistics is a low-margin, high-coordination physical industry. It fits Kass's thesis of "Real World" sectors benefiting from efficiency gains provided by AI, unlike pure software companies which are threatened by it. LONG. Transportation and logistics are prime beneficiaries of algorithmic efficiency. Rising fuel costs or economic slowdown reducing freight volume.