Eclipse Ventures Nets $2.5B from Cerebras IPO After 2016 Series A Bet
Lior Susan started Eclipse Ventures in 2015 with a lonely thesis: that digitizing the physical world — chips, robots, factories, space — would print mon...
Lior Susan started Eclipse Ventures in 2015 with a lonely thesis: that digitizing the physical world — chips, robots, factories, space — would print money while everyone else chased SaaS unicorns. That bet just paid off to the tune of $2.5 billion when Cerebras Systems went public this week, delivering Eclipse a 17x return on its $147 million total investment. For AI tool builders, newsletter writers, and startup watchers in Delhi, this isn't just a VC victory lap — it's a flashing neon sign that the easy money in pure software is gone, and the next fortune is being forged in silicon, steel, and clean rooms.
What Is Eclipse Ventures and Its "Physical World" Thesis?
Eclipse Ventures was founded in 2015 by Lior Susan with a focus on investing in companies that build the underlying infrastructure of the physical economy — not just software that runs on top of it. At the time, Silicon Valley was obsessed with SaaS, mobile apps, and cloud platforms. Susan’s bet? That 85% of global GDP is tied to the physical world (manufacturing, energy, transport, defense), and that digitizing those sectors would require hardware, semiconductors, robotics, and deep-tech engineering — not just a subscription model.
- Core sectors: Semiconductors, robotics, energy, space, defense, industrial automation.
- Key distinction: These companies require capital-intensive R&D, long development cycles, and deep supply chain expertise — not just "vibe coding" an MVP over a weekend.
- Why it matters now: AI’s insatiable hunger for compute (chips, data centers) and its ability to finally make robotics viable have turned these once-unfashionable sectors into the hottest tickets in venture.
The Core News: Cerebras IPO Delivers a $2.5 Billion Windfall
Cerebras Systems, the AI chip company known for building the world’s largest processor (the Wafer Scale Engine), went public this week at $185 per share. Eclipse Ventures was the Series A lead back in 2016 with a $6.5 million check, and over time invested a total of $147 million. At the IPO price, that stake is now worth $2.5 billion — a 17x return.
| Stage | Eclipse Investment | Today’s Value at IPO |
|---|---|---|
| Series A (2016) | $6.5M lead | $2.5B total return |
| Total over time | $147M (multiple rounds) | 17x multiple on total cost |
- This is not just a single lucky bet: Eclipse also saw $4.5 billion in follow-on rounds across its portfolio in Q1 2026 alone — up from less than $4 billion in total during its first eight years.
- Notable recent rounds: $1.2 billion for Wayve (AI for autonomous driving), $650 million for True Anomaly (space security), $270 million for Bedrock Robotics (underground construction robots), and $200 million for Oxide Computer (cloud-optimized hardware).
- All four of those companies were Series A investments by Eclipse — a track record most VC firms can only dream of.
Susan’s blunt explanation for the shift: "The real moat in software is gone. You can vibe code pretty much whatever you want." With AI tools like Anthropic’s Claude Code and OpenAI’s latest models, any enterprise can generate custom software in-house. But you can’t vibe code a chip fab.
Why This Matters: The "Software Moat Is Dead" Thesis
This is the paragraph where the implications hit home for every founder, marketer, and AI tool publisher reading in Delhi. Susan’s argument — that software alone no longer offers defensibility — is a direct threat to the SaaS-first mindset that has dominated Indian tech startups for a decade.
- "Vibe coding" (using AI to generate software from natural language prompts) means any business can create its own tools. Why pay for a SaaS CRM when your intern can ask GPT-6 to build one overnight?
- The result? SaaS stocks tumbled in early 2025 on fears that enterprise AI would eat recurring revenue.
- Hardware and physical-world infrastructure can’t be replicated by AI prompts. You need patents, supply chains, clean rooms, and government subsidies — all barriers to entry that create durable moats.
- For Indian founders, this is a wake-up call: building yet another B2B SaaS wrapper on OpenAI’s API is a race to zero. The real opportunity lies in hardware-adjacent AI — chips, robotics, drone logistics, energy storage, and industrial automation.
| Industry | Software-Only Play | Physical-World Play |
|---|---|---|
| AI Assistants | Chatbot wrapper (easily copied) | Custom silicon for inference (Cerebras) |
| Logistics | Last-mile routing app (vibe-codable) | Autonomous delivery robots (Bedrock) |
| Space | Satellite data analytics (SaaS) | Manufacturing satellites and launch systems (True Anomaly) |
| Energy | Solar panel monitoring dashboard | Grid-scale battery storage + AI control |
Key Details: The Five Forces Behind the Shift
Susan argues that five forces are now aligned for the first time in American industrial history (since Henry Ford and Carnegie) to make physical-world tech the dominant venture category.
Technology
- AI is the catalyst: compute demand drives chip investment, and AI algorithms finally enable robots to navigate messy physical environments.
- Generative AI also makes software cheaper, pushing capital toward hardware moats.
Capital
- Late-stage VC rounds are flooding into physical-world companies — $15 billion in Eclipse’s own portfolio in 2025 alone.
- Public markets are rewarding hardware: TSMC and Micron hit all-time highs recently.
Customer Demand
- Every large enterprise needs custom silicon, robotic automation, and secure supply chains.
- Governments are the biggest customers for defense and space tech.
Talent
- Top engineers are leaving SaaS to build robots, rockets, and chips. They’re tired of A/B testing button colors; they want to build things that move and grind.
Policy
- The U.S. government is actively subsidizing semiconductor fabs (CHIPS Act), space R&D, and defense tech. Favorable regulation and tax incentives are accelerating deployment.
Competitive Landscape / Industry Context
Eclipse is not alone, but it has the earliest and deepest commitment to this thesis among venture firms.
- Andreessen Horowitz (a16z) also has a massive hardware practice (e.g., Anduril, SpaceX investments), but its portfolio is broader.
- Sequoia Capital has backed semiconductor firms like Nvidia and Arm, but its focus is still heavily weighted toward software.
- Tiger Global and SoftBank poured billions into late-stage hardware companies (e.g., Arm, OpenAI’s compute infrastructure), but they lack the early-stage, multi-sector physical-world specialization Eclipse claims.
Eclipse’s differentiator: it was the Series A investor in most of its breakout successes, and it has been patient — waiting a decade for the thesis to mature. This is the opposite of the fast-flip SaaS model.
What This Means for AI-Tool and AI-News Publishers
This story is a goldmine for content creators, AI newsletter writers, and SEO bloggers focused on the Indian market. Here are five concrete content angles you can use right now:
-
"Vibe Coding Is Eating SaaS — Here’s What Physical-World Startups Indian Founders Should Build Instead"
Write a practical guide for Indian tech entrepreneurs on hardware-adjacent opportunities: custom ASICs for inference, agricultural robotics, drone manufacturing for last-mile delivery, solar panel cleaning robots, and industrial IoT. -
"Cerebras IPO: What It Teaches Us About the Future of AI Compute"
Break down why wafer-scale chips matter for AI training, and how this affects cloud pricing for Indian startups using GPT alternatives. -
"The Five Forces Behind the ‘Physical World’ VC Boom — And Why India Should Pay Attention"
Adapt Susan’s framework for the Indian context: government incentives (like PLI for semiconductors), talent moving from IT services to hardware, and rising demand from defense and telecom. -
"How AI Is Reshaping the VC Playbook: From SaaS Multiples to Hardware Moats"
Compare valuation metrics: SaaS companies trade on ARR growth, but hardware startups trade on capital efficiency, patent portfolios, and government contracts. Ideal for your finance-savvy readers. -
"Top 10 Indian Startups Building Real-World AI (Not Just Another Chatbot)"
Curate a list of Indian deep-tech startups in chips (e.g., Mindgrove), robotics (e.g., GenRobotics), energy (e.g., Log9), and defense (e.g., Tonbo Imaging). Pitch this as a “where to invest your time/money” guide.
SEO opportunities: Target keywords like “physical world investing,” “vibe coding meaning,” “Cerebras IPO impact,” “hardware vs software moat,” “Indian semiconductor startups 2026,” and “AI robotics venture capital.”
Challenges Ahead / Risks / Limitations
- Capital intensity: Physical-world startups need 10x more capital than SaaS to reach scale. A single wafer fab costs billions. This limits the investable universe.
- Long time to exit: Eclipse waited 10+ years for Cerebras. Many VCs don’t have that patience.
- Geopolitical risk: Semiconductor investments are heavily dependent on U.S.-China tensions and export controls. A policy flip could devastate portfolios.
- Execution difficulty: Hardware is hard. Supply chains break. Prototypes fail. The failure rate is higher than software.
- Talent shortage: India has a vast pool of software engineers but a thin pipeline of hardware engineers and chip designers. Building an ecosystem takes decades.
- Overvaluation risk: The rush into physical-world tech could create bubbles — just as we saw with SaaS during ZIRP (zero interest rate policy). Not every robotics startup is the next Wayve.
Final Thoughts
Eclipse’s $2.5 billion Cerebras win is a trophy case proof-point that hardware and physical infrastructure are the new venture capital darlings — and that the era of easy software wealth is fading. For Indian founders and creators, the lesson is clear: look beyond the “next Uber for X” and toward the machines, chips, and robots that will power the AI economy. The moat you can’t vibe code is the moat that endures.
FAQ
What is the "physical world" investing thesis of Eclipse Ventures?
Eclipse focuses on companies that digitize and build the infrastructure of the real economy — semiconductors, robotics, energy, defense, and space — rather than pure software.
How much did Eclipse make from the Cerebras IPO?
Eclipse invested a total of $147 million across multiple rounds in Cerebras. At the IPO price of $185 per share, that stake is worth $2.5 billion, generating a 17x return.
What does "vibe coding" mean, and why is it relevant?
"Vibe coding" is the practice of using AI tools like Claude Code to generate software from natural language prompts. Lior Susan argues this makes software easy and copyable, eroding the defensibility of pure software companies.
Which other companies has Eclipse backed that are now hot?
Key portfolio companies include Wayve (autonomous driving AI, $1.2B raise), True Anomaly (space security, $650M), Bedrock Robotics (underground robots, $270M), and Oxide Computer (cloud hardware, $200M).
Is this a US-only trend, or does it affect India?
It directly affects India. The same dynamics — AI making software cheap, government incentives for hardware, and talent shifting from SaaS to deep tech — are accelerating in India, especially in semiconductors (PLI scheme), robotics, and defense.
What are the risks for investors in this space?
High capital intensity, long development timelines, geopolitical risks in chip supply chains, and potential overvaluation as money floods into physical-world startups. Not every hardware bet will succeed.
