Why Applied Deep Tech Beats General AI for B2B Investors Right Now

The market is rewarding AI exposure. It is not yet rewarding AI durability. That distinction matters. In the last two years, enterprise AI adoption has expanded quickly, but the economic picture remains uneven. McKinsey’s 2025 global survey found that 88% of organizations report regular AI use in at least one business function, yet only about one-third say they have begun to scale AI across the enterprise, and just 39% report EBIT impact at the enterprise level. In other words, usage is broad, but durable value creation is still narrow.

For B2B investors, that gap is the signal. General AI is becoming infrastructure. Applied deep tech is where the moat is being built. At Azafran Capital Partners, that is not a semantic distinction. It is an investment posture. We focus on applied deep tech because the next decade of outperformance in MedTech, IoT, and enterprise B2B will not come from the biggest model. It will come from defensible intellectual property, domain specificity, and operational systems that solve costly real-world problems inside complex industries. 

General AI is scaling fast, but advantage is compressing 

The first wave of AI investing rewarded access: access to models, access to talent, access to customer attention. That window is narrowing. Enterprise buyers are already becoming more sophisticated in how they purchase and deploy AI. Andreessen Horowitz’s 2025 CIO survey found that organizations are increasingly mixing and matching multiple models to optimize for performance and cost rather than committing to a single model stack. That is rational buyer behavior, but it also means the foundational model layer is becoming more substitutable from the standpoint of many B2B use cases.

McKinsey’s findings reinforce the same point from the operator side. The companies seeing the most value are not simply “using AI.” They are redesigning workflows, embedding governance, and aligning operating models, data, and delivery around measurable business outcomes. High performers are nearly three times as likely as others to have fundamentally redesigned workflows. That is why we believe general AI, by itself, is not the investable thesis for most B2B companies. It is an input. A powerful one, yes. But still an input. The investable thesis is what a company can uniquely do with that input in a regulated workflow, a mission-critical operating environment, or a technical system where performance, explainability, and integration matter more than novelty. 

Applied deep tech creates moats where B2B buyers actually care 

In enterprise markets, buyers do not pay a premium for abstraction. They pay for outcomes they can validate, deploy, govern, and defend. 

That is where applied deep tech separates from general AI. Applied deep tech combines advanced capabilities such as voice, acoustics, imagery, machine learning, and augmented AI with domain-specific workflows and, critically, protectable technical differentiation. It is closer to systems engineering than software theater. 

This matters more in sectors like MedTech and IoT because those markets punish superficial technology. Clinical environments, connected devices, industrial operations, and enterprise infrastructure all demand precision, reliability, interoperability, and regulatory discipline. A horizontal tool may demo well. It often fails at deployment. 

The history of IoT platforms is a useful warning. IoT Analytics noted in 2025 that the long-promised horizontal IoT platform opportunity had turned from a “blue ocean” into a “red lake,” precisely because deployment complexity proved harder than expected and many end users did not see enough tangible value from generic platforms.

That pattern repeats across B2B. Generic capability attracts attention. Domain-engineered capability wins budgets. For investors, the lesson is straightforward: the more a solution is embedded in technical workflow, regulatory context, and operational data, the harder it is to replace. That is what durable enterprise value looks like. 

Defensible intellectual property matters more now, not less 

A surprising number of investors talk about AI as though IP no longer matters because models are advancing so quickly. We take the opposite view. As foundation models become more accessible, the scarce asset shifts upstream and downstream: proprietary data pipelines, integration architecture, workflow design, and defensible intellectual property around the actual application of the technology. 

That is not theory. Global patent activity continues to rise. WIPO reported a record 3.7 million patent applications filed worldwide in 2024, up 4.9% year over year, with computer technology the largest category and medical technology also among the top fields in published applications. WIPO also notes that foreign-oriented patent families are generally considered higher quality and higher value because filing across jurisdictions is costly and selective.

The legal environment is also evolving in a way that reinforces the value of human-led invention. In late 2025, the USPTO issued revised guidance confirming that the same inventorship standard applies to AI-assisted inventions as to any other invention. That matters because it preserves the importance of actual inventive contribution rather than reducing innovation to model output.

For B2B investors, that means defensible IP is not an old-world artifact. It is one of the clearest indicators that a company is building something more durable than an interface on top of a commoditizing stack. 

In MedTech and industrial systems, the bar is higher — and that is attractive 

We like markets where the bar is high. In MedTech, complexity filters noise. EY’s 2025 Pulse of the MedTech Report found that venture activity remained particularly robust, with average financing rounds reaching $36 million in the first half of 2025, up 122% over 2024. That is not capital chasing hype. That is capital concentrating around companies with stronger evidence, clearer pathways, and more credible technical differentiation.

The same logic applies to enterprise and industrial environments. Buyers in these sectors care about uptime, risk, compliance, signal quality, and integration with the systems already running the business. They do not want “AI.” They want a measurable improvement in clinical workflow, operating efficiency, decision support, device performance, or customer economics. This is why applied deep tech often produces better investment geometry than general AI in B2B. The product is harder to build, the sales cycle is more demanding, and the diligence burden is heavier. But once adopted, the solution tends to be more embedded, more defensible, and more valuable. In our view, that is exactly where investors should want to work. 

Capital is not enough; operational acceleration is part of the moat 

One more point is often missed in venture discussions: deep tech returns are not created by capital alone. BCG has argued that deep tech has matured into an established asset class, holding roughly 20% of venture funding since 2019 even through broader market resets. The same research also highlights that deep tech investors rely heavily on technical expertise and external specialists to evaluate and support companies. 

We agree with the premise, but we would take it further. In applied deep tech, value accretion comes from operational excellence, not financial engineering. 

That is the logic behind the Azafran Catalyst. We do not view ourselves as transactional capital. We are long-term partners who help companies move from technical promise to market success through operating discipline, commercial clarity, and what we call the Principles-First Thinking Framework. In practice, that means helping founders align product, governance, go-to-market execution, and scaling decisions before complexity compounds. The best applied deep tech companies are not merely inventing new capabilities. They are building systems that can survive diligence, deployment, procurement, and scale. 

That requires more than a term sheet. 

The investment posture from here 

We are bullish on AI. We are more bullish on applied deep tech. General AI will continue to create enormous value, but much of that value will accrue to infrastructure providers and a smaller number of scaled platforms. For most B2B investors, the more compelling opportunity is downstream: companies that apply advanced technologies to hard industry problems, protect that work with defensible intellectual property, and build real operating leverage around adoption. 

That is especially true in MedTech, IoT, and enterprise B2B, where domain specificity is not a feature. It is the moat. Our view is simple: in this market, the winners will not be the companies with the loudest AI narrative. They will be the companies with the strongest technical differentiation, the clearest operational discipline, and the deepest fit with the markets they serve. That is why applied deep tech beats general AI for B2B investors right now. And from where we sit, that will only become more true over the next cycle.