**Author’s note: The following post should be read as a “thesis in development”. One of the most wonderful things about working in venture capital is having the mandate to let our minds wander and explore various sectors and strategies. At TLV Partners we remain primarily focused on software-based investments, but due to the reasons outlined below, are more open to considering hardware related opportunities than in the past.**
A Quick History Lesson
In many ways, the entire high-tech world is built on a foundation of hardware. The very name “Silicon Valley” traces back to the semiconductor industry—companies like Fairchild Semiconductor, Intel, AMD and others that formed the bedrock of modern computing. Without the chips that power the world as we know it today, there’d be no platform on which the software industry would exist.
In fact, venture capital itself was born out of funding these early hardware ventures. (For a truly fantastic quick history on the origins of VC and its hardware roots, this video is a must watch).
But Microsoft’s dominance in the 1980s and the consumerization of the internet in the 1990s shifted venture capitalists’ focus from hardware to software. And rightly so. VC’s evolution from a cottage industry to an asset class can be largely attributed to the economic windfall that software created.
By the time I started my VC career in 2014, the conventional wisdom rang out loud and clear: hardware is too risky; it’s less scalable; margins are too low—software is the land of opportunity.
I think this conventional wisdom may need an update.
The Shifting Sands of Software
Over the last decade, software’s dominance has seemed almost unstoppable. But there’s an important shift happening right now: the proliferation and rapid evolution of AI. LLMs and the products built using them have proved that software is going through a process of commoditization.
More recent advancements, including the fusing of reinforcement learning, transformers and chain of thought reasoning, further demonstrate that a future in which software talent is no longer a constraint for building technological economic value is closer than expected.
If AI continues to advance at its current pace, software that once required entire teams of engineers over many months (if not years) to build may soon be built, or significantly augmented, by AI in a fraction of that time.
That scenario raises a critical question: If software’s barrier to entry shrinks, where can defensible moats be found?
The Value of AI-Driven Physical Products
Today, most hardware components themselves are commodities. But building out and optimizing a supply chain, manufacturing and logistics are complex. They’re a schlep. Most people simply don’t have the stomach for it.
Companies that are able to figure out the difficulties surrounding building a physical product can have inherent moats simply based on the fact that figuring hardware out is a moat in and of itself.
Beyond that, switching costs for a physical product are often higher than for purely digital products. Consumers and businesses get emotionally attached to products that bring them joy, utility, or status—capturing that element can be priceless, and defensible.
It’s no surprise then, that four of the top ten companies in the world by market cap today—Apple, NVIDIA, Tesla, and TSMC—are deeply rooted in hardware.
Those companies were founded decades ago though.
But what happens when the challenges surrounding hardware are married with the benefits of AI?
I imagine a future in which physical products are used as a wedge offering in order to land customers (both consumers and enterprise), while the true value is derived from an associated and ongoing AI service.
Furthermore, AI can be used in order to help solve many of the challenges related to supply chain management and manufacturing. We’ve met a long list of “AI for supply chain” related companies, dealing with everything from component sourcing agents, to BPO automation and everything in between.
So yes, hardware margins may be lower than the canonical 85%+ software margins. But let’s not forget that when you build true moats, you can command pricing power. Just look at NVIDIA to see how a hardware-centric business can thrive.
It seems to me that the real opportunity in 2025 and beyond lies at the intersection of AI and hardware—where software intelligence meets physical innovation.
Why Israel Is Poised to Lead
So, how does Israel fit into all this? The country has a deep bench of hardware talent, much of it stemming from the military.
Beyond the military, Israel’s hardware track record includes powerhouses like Mobileye, Mellanox, SolarEdge, and Habana Labs. Global giants such as Intel, Nvidia, and Amazon have significant semiconductor R&D centers in Israel for a reason.
Double clicking on this point, 2025 represents an interesting point in time for many of the companies listed above.
- Intel’s local talent pool is either actively pursuing new opportunities or can be easily swayed to work on a startup due to the uncertainty regarding the future of the company.
- Habana Labs’ acquisition was over four years ago. And despite its success within Intel, this is usually the time when talent begins to explore new opportunities.
- Mobileye has experienced turbulence post its independence in the public markets.
- Mellanox/Nvidia seems to be doubling down on Israel, and that’s a great thing for the country! And until recently, turning down Nvidia RSUs was an almost incomprehensible decision. But perhaps Deepseek has made a chink in the green armor.
- SolarEdge is a fantastic company and has been the role model of a truly built-in-Israel company in the public markets, but it has experienced significant turbulence over the past six months.
Add to the above list smaller companies that have achieved varying levels of success and have experienced what it means to build hardware (Innoviz, Hailo, Next Silicon, Vayyar, Storedot, Arbe Robotics, etc…) and there should be a dearth of experienced hardware talent in the marketplace. *Interestingly, there haven’t (yet) been any prominent Paypal Mafia-esque groupings that have emerged in Israel’s hardware scene.*
Combine that hardware talent with the brilliant software talent the ecosystem has developed over the past two decades, as well as the constant progress in AI that we’re experiencing as we speak, and you get a potent recipe for a new wave of AI-powered hardware companies.
There are several categories where this fusion of hardware and AI could unlock massive value:
AI Native Tools That Support Physical World Endeavors
These are companies that help speed up the current hardware development process. This could be anything from AI-based procurement, to AI-based EDA tools, AI-enabled manufacturing and more.
While I have argued above that we are at the precipice of a hardware renaissance, Israel has been historically challenged when it comes to manufacturing. I’m not sure that the next great physical product company will emerge from Israel, but I am confident that Israel has the talent to support the next generation of important hardware companies built across the world.
Next-Gen Compute
During the previous AI wave, primarily surrounding computer vision back in 2017, a great cohort of Israeli next-gen computer companies were founded. The aforementioned Habana Labs, Next Silicon, Hailo, Neuroblade and more. In addition to even further optimized chipsets for LLMs, I believe there are opportunities at the networking layer as well as further exploring different platforms altogether such as Quantum Computing.
Defense
We’ve previously written about the opportunity for a new wave of defense primes.
Consumer Hardware
Perhaps the first example of AI penetrating mainstream households was Alexa. What other consumer devices will consumers be able to interact with in the future, and what new products will be introduced into the home/workplace/public space that become as habitual as asking Alexa to set a timer or help with measurement conversions?
Robotics
Rapid advancements in robotic system-centric foundation models makes the inevitability of a robot infused world clearer and clearer by the day. Companies like Figure, Physical Intelligence and Skild are some of the leaders in the digital building blocks behind the robotic revolution, and companies like Unitree (and hopefully some Western upstarts) are leading the way from a physical build out perspective. We believe that there continues to be an opportunity to be an integrator of the best of breed AI and hardware in order to create purpose built robotics for various industries.
Conclusion: Betting on a Hardware Future
Venture capital began with hardware, drifted toward software, and is now at a point where both are critical. As AI makes pure software increasingly commoditized, hardware offers a tangible, defensible layer that can propel startups to outlier success.
There’s no denying the challenges—supply chain complexities, lower margins, and longer development cycles—but these very hurdles can be transformed into competitive advantages if approached with the right mindset (and a strong stomach). Given Israel’s wealth of hardware talent, combined with its renowned software expertise, I believe we’re on the brink of a hardware renaissance—one powered by the latest advances in AI.
If you’re a founder in this AI-meets-hardware domain, I’d love to hear from you. Let’s build the next generation of transformative tech—both the bits and the atoms.
The post Hardware Strikes Back: Investing in AI-Driven Hardware appeared first on TLV Partners.