Eclipse’s Cerebras Windfall Signals a Broader Shift Toward ‘Physical World’ Tech

A long-running venture thesis is suddenly in favor

Eclipse Ventures is using the public-market success of Cerebras Systems to argue that Silicon Valley’s center of gravity is moving back toward technology rooted in the physical world. TechCrunch reports that Eclipse’s initial $6.5 million Series A investment in Cerebras in 2016 ultimately contributed to a $2.5 billion return after the semiconductor company went public this week. Over time, the firm invested a total of $147 million in Cerebras, producing what it says is a 17-fold return at the IPO price of $185 a share.

That result is notable on its own. But the deeper significance lies in what Eclipse founder Lior Susan says it represents: a validation of the idea that companies operating at the intersection of hardware and software can generate some of the biggest outcomes in tech. In a period when frontier AI has made pure software easier to create and faster to imitate, scarcity may be shifting toward the systems that touch materials, machines, energy, mobility, and defense.

Why the “physical world” thesis is resonating now

Susan’s argument, as described by TechCrunch, is straightforward. Roughly 85 percent of global GDP is tied to the physical world. If software’s moat is weakening because advanced AI tools can help many teams build products faster, then investors may start placing higher value on companies that still require manufacturing, supply chains, specialized infrastructure, and difficult operational execution. In that logic, chips, robotics, and energy systems gain strategic importance precisely because they cannot be conjured into existence through software abstraction alone.

The source captures this view in especially clear terms when discussing the limits of “vibe coding.” Software can be created more easily than before, but semiconductor manufacturing still depends on silicon, machines, clean rooms, and industrial process know-how. That makes the physical layer harder to copy and potentially more defensible.

Cerebras as proof point

Cerebras is a fitting example because it sits at the intersection of AI demand and hardware difficulty. AI growth has made compute infrastructure one of the most valuable choke points in the technology economy. A company that can build differentiated semiconductor systems stands to benefit not only from software demand but from the fact that modern AI increasingly depends on specialized physical infrastructure.

From Eclipse’s perspective, the IPO outcome does more than validate one bet. It helps justify a portfolio strategy that for years looked out of step with a venture market heavily oriented toward software and SaaS. Susan told TechCrunch that investing in the physical world felt lonely in the early years of the firm. Now, public-market moves in names like TSMC and Micron, along with the fundraising pace of hardware-linked startups, suggest the thesis has become mainstream.

The rest of the portfolio tells the same story

TechCrunch says Eclipse portfolio companies in sectors including robotics, energy, and defense raised nearly $15 billion from outside backers last year, with another $4.5 billion arriving in the first quarter of 2026 alone. The article cites major rounds involving Wayve, True Anomaly, Bedrock Robotics, and Oxide Computer, all companies where Eclipse was the Series A investor.

That matters because it broadens the argument beyond semiconductors. The new enthusiasm is not limited to chips. It reflects a wider belief that high-value innovation increasingly lives where computation meets the physical world: autonomous systems, industrial infrastructure, defense technology, and energy platforms that are expensive to build but hard to displace once they work.

What this means for innovation markets

• AI may be compressing the defensibility of some pure-software categories.
• Investors are responding by favoring companies with deeper physical and industrial moats.
• Semiconductors, robotics, energy, and defense are benefiting from that shift.
• Public-market outcomes like Cerebras give venture firms stronger evidence to keep backing capital-intensive bets.

The innovation takeaway is not that software stops mattering. It is that the market may be repricing where durable advantage lives. If AI makes code easier to generate, then factories, chips, robots, and infrastructure become relatively more strategic, not less. Eclipse’s Cerebras return is therefore more than a venture victory lap. It is a signal that the next technology cycle may reward companies that can move bits through atoms, not just screens.

This article is based on reporting by TechCrunch. Read the original article.