AI's Profitability Problem: Compute Costs Versus Revenue Uncertainty

OpenAI's revenue miss and the subsequent market jitters highlight a fundamental tension in the AI gold rush: the immense cost of innovation versus the uncertain path to profitability. This conversation reveals that while the allure of cutting-edge AI is strong, the underlying economics are far from settled. Investors and stakeholders who hitched their wagons to OpenAI's star, like Oracle and CoreWeave, are now facing the immediate consequences of this uncertainty. Anyone investing in or relying on the AI infrastructure boom should read this to understand the hidden costs and long-term viability challenges that conventional wisdom often overlooks, gaining an advantage by preparing for a more complex and potentially less lucrative future than currently advertised.

The Compute Chokehold: Why AI's Profitability Problem Is Deeper Than You Think

The recent news that OpenAI missed internal growth and revenue targets sent ripples through the tech sector, particularly hitting companies with significant ties to the AI giant. This isn't just a minor setback; it exposes a critical systemic flaw: the staggering cost of AI development and deployment. While the capabilities of models like ChatGPT capture the public imagination, the underlying economics are a stark reality check. The conversation on Motley Fool Hidden Gems Investing dives deep into this, revealing that the path to profitability for AI companies, and by extension their partners, is far more precarious than often portrayed.

The core issue, as articulated by the hosts, is the immense spending required for compute power. This isn't a problem that’s going away; it’s a fundamental constraint. Lou Whiteman points out the stark reality:

"My question is, how does any of this make sense, guys? Who's going to make money here? Part of the reason OpenAI put out outrageous revenue assumptions is they have to offset outrageous spending needs. Anthropic is throttling people because compute is so expensive. OpenAI still needs to raise money. There is, I think the market just needs to wake up to just how much money is needed here."

This highlights a cascading effect. OpenAI’s ambitious goals necessitate massive investment in compute, which in turn requires equally ambitious revenue targets. When those targets are missed, the financial strain is felt not only by OpenAI but also by its partners who have made substantial commitments. Oracle, for instance, has seen its stock decline significantly since announcing its partnership with OpenAI, a clear indication of investor skepticism about the long-term feasibility and profitability of these AI deals. This isn't just about one company missing a target; it's about the entire ecosystem built around these high-cost models facing a reckoning.

The Illusion of First-Mover Advantage

In the fast-paced world of AI, the narrative often centers on first-mover advantage. However, the podcast discussion suggests this may be a misleading metric. Lou Whiteman argues that the market is still forming, and history shows that dominance can shift rapidly. The rise of Google in internet search, displacing earlier giants like Netscape and Ask Jeeves, serves as a potent historical parallel. This implies that even if OpenAI falters, a new player could emerge, rendering current investments and strategies obsolete. The implication for investors and businesses is that betting solely on current leaders might be a precarious strategy in a market characterized by rapid disruption.

This dynamic is further underscored by the emergence of more cost-efficient AI models. The development of open-source models like DeepSeek, which claim to achieve comparable performance at a fraction of the compute cost, directly challenges the economic assumptions of closed-loop, high-cost models. Matt Frankel notes this shift:

"I think for a while, the conversation around AI has been capability, you know, that real wow factor of like what it can do... But I think we're going to now start seeing with these LLM models a focus more on cost efficiency. It's going to be a part of the conversation because as you said, right now, no one's making money with this, and eventually creditors, investors, they will want to see something that's moving towards something that doesn't look like a vacuum sucking every dollar out of your wallet, or at least something that can, you know, cover expenses."

This focus on cost efficiency is a crucial downstream effect. If models become significantly cheaper to run, the economic viability of AI services changes dramatically. It could reduce the pressure on companies like OpenAI to achieve astronomical revenue targets and, conversely, empower smaller players or those who can leverage these cheaper models. The competitive landscape is not just about who has the most powerful model, but who can deliver it sustainably and profitably.

The Auto Industry's Software Pivot: A Risky Bet on High Margins

Shifting gears to General Motors (GM), the discussion highlights a different kind of economic challenge: the transition to electric vehicles (EVs) and the increasing reliance on software and services for profitability. While GM posted better-than-expected earnings, the underlying narrative is one of navigating a brutal, cyclical industry while attempting to capture the high margins associated with software. Matt Frankel champions GM, pointing to its strong market share and the growth in its software offerings like Super Cruise and OnStar.

However, the podcast raises a critical question about the sustainability of these high-margin software revenues. Tyler Crowe expresses skepticism, drawing a parallel to how premium features in the auto industry historically become standard over time.

"I'm really curious to see how sustainable that is because you have a hundred-year tradition in the auto business of features starting as premium and moving downstream to standard. I mean, my Honda can do 90% of what Super Cruise does, and it came as standard, not non-subscription."

This points to a delayed payoff scenario. GM is investing heavily in its EV strategy and its software ecosystem, hoping these will provide a durable competitive advantage and a stable, high-margin revenue stream. The immediate cost of this pivot, including one-time charges related to strategy shifts, is significant. The long-term success hinges on whether consumers will continue to pay subscription fees for features that might eventually become commoditized or integrated into standard vehicle offerings. If this software revenue doesn't materialize as expected, GM’s core business, which is inherently low-margin and highly competitive, will bear the brunt. The hope is that the software side will offset the cyclicality and razor-thin margins of traditional auto manufacturing, creating a more resilient business model. But this is a bet on future consumer behavior and technological adoption that carries inherent risk.

The Power of the Proxy: Why Your Vote Matters (Even If It Feels Small)

The final segment addresses a mailbag question about proxy voting for individual shareholders. While Matt Frankel and Lou Whiteman admit to often neglecting their votes due to practicalities and a perceived lack of individual impact, Tyler Crowe passionately argues for the importance of shareholder participation. He frames it not just as a right but as a duty for those who choose to invest in individual companies rather than diversified funds.

"If you own individual stocks, you should care. You should read your proxy filings. You should vote on everything that you have... management works for us, and it's our duty to vote on the results of this business and how and the things that they're asking us to do."

This is a clear example of where immediate discomfort (the time and effort required to vote) can lead to a long-term advantage for the system as a whole. By actively participating, shareholders can hold management accountable, influence corporate governance, and ensure that executive compensation aligns with shareholder interests. While an individual vote might seem insignificant in a large corporation, collective action can drive change. The "discomfort now" comes from the effort of engagement, but the "advantage later" is a more responsibly managed company, potentially leading to better long-term investment performance. This insight underscores the idea that even in financial markets, active participation, though often overlooked, is a powerful lever for systemic improvement.

• Immediate Action (Within the next quarter):

  • AI Investment Due Diligence: Scrutinize the compute costs and revenue models of AI companies and their partners. Prioritize businesses demonstrating a clear path to profitability beyond speculative growth.
  • Evaluate Auto Software Subscriptions: For investors in automakers like GM, assess the long-term sustainability of software subscription revenue versus the commoditization of features.
  • Review Proxy Materials: If you own individual stocks, commit to reading proxy statements and voting on all proposals, especially those related to executive compensation and board elections.

• Longer-Term Investments (6-18 months and beyond):

  • Diversify AI Exposure: Consider investments in companies focused on AI efficiency and cost reduction, not just those building the most powerful models.
  • Monitor Auto Industry Margins: Track how software revenue contributes to overall profitability for automakers and watch for trends in feature commoditization.
  • Build a Governance Framework: For active investors, develop a consistent process for researching and voting proxies to ensure ongoing accountability.
  • Seek Durable Competitive Advantages: In both AI and automotive sectors, look for businesses that are building moats not just through innovation, but through sustainable economic models that can withstand market shifts and competitive pressures. This requires patience, as these advantages often take years to fully develop.