Ten years ago, when Lior Susan started Eclipse Ventures, his focus on startups solving real-world problems with hardware felt risky. Most venture capitalists chased software or consumer apps. Susan bet on robots, industrial AI, and chips—not the flashy SaaS models that dominated Silicon Valley. By 2024, his gamble paid off in a big way: Eclipse led a $2.5 billion round for Cerebras Systems, the AI chipmaker building the hardware behind today’s most demanding AI workloads. Cerebras isn’t just another startup—it’s making the silicon that runs the AI models everyone’s talking about, from climate simulations to drug discovery. The deal proves real-world AI investing isn’t a niche anymore. It’s where the smart money’s going.

Cerebras’s breakthrough is its wafer-scale chip, a single silicon wafer packed with 2.6 trillion transistors. That’s 500 times more than a standard GPU, and it lets Cerebras crunch AI models at speeds no one else can match. Companies like G42 in the UAE and Argonne National Laboratory in Illinois are already using Cerebras chips to train massive models or run climate simulations. The $2.5 billion Eclipse invested will help Cerebras scale up manufacturing, hire talent, and push further into industries where speed and precision matter more than cost.

Silicon Valley finally agrees with Susan’s vision

Back in 2014, when Eclipse launched, most VCs ignored hardware. Today, the tables have turned. Nvidia’s dominance in AI chips has shown Silicon Valley the power of owning the physical stuff that runs AI. Cerebras is one of the few companies trying to compete directly with Nvidia in the high-end AI chip market. Its approach—big, powerful chips designed for specific workloads—isn’t just different; it’s necessary. While Nvidia’s GPUs are versatile, Cerebras’s chips excel at running massive AI models without the bottlenecks that slow down traditional systems.

Eclipse’s bet wasn’t just about money. The firm has spent years building relationships with engineers and researchers who need next-gen hardware. Susan’s team knew Cerebras’s chips could solve problems in energy, biotech, and national security—areas where software alone falls short. The $2.5 billion round signals that investors now see those problems as critical, not experimental.

The real-world impact isn’t just hype

Cerebras’s chips are already at work in places you wouldn’t expect. At Argonne National Laboratory, researchers use them to model nuclear fusion reactions, a task that would take months on traditional supercomputers. In Abu Dhabi, G42 is training large language models with Cerebras chips to support Arabic-language AI tools. These aren’t lab experiments—they’re real deployments proving the tech works. For Eclipse, the Cerebras deal is validation that hardware-focused AI startups can scale and deliver results.

Susan isn’t surprised by the shift. He’s been saying for years that AI’s next big breakthroughs would come from better hardware, not just bigger models. The Cerebras investment is proof he was right. It’s also a wake-up call to other VCs who’ve been slow to bet on the physical side of AI. If Eclipse can turn $2.5 billion into a portfolio of successful companies, others will follow.

What happens next? Cerebras plans to double down on manufacturing and expand into new markets like healthcare and autonomous systems. Eclipse, meanwhile, is already scouting the next big hardware bet. With Nvidia’s grip on AI chips tightening, the race is on to find alternatives that can keep up with the demand for faster, more efficient computing. The Cerebras deal isn’t just a win for Eclipse—it’s a signal that the tech world’s next big wave isn’t just in the cloud. It’s in the machines.

What You Need to Know

  • Source: TechCrunch
  • Published: May 17, 2026 at 15:00 UTC
  • Category: Technology
  • Topics: #techcrunch · #startups · #tech · #for-eclipse · #cerebras · #investing

Read the Full Story

This is a curated summary. For the complete article, original data, quotes and full analysis:

Read the full story on TechCrunch →

All reporting rights belong to the respective author(s) at TechCrunch. GlobalBR News summarizes publicly available content to help readers discover the most relevant global news.

Curated by GlobalBR News · May 17, 2026

🇧🇷 Resumo em Português

O Eclipse, fundo de investimentos bilionário, acaba de abocanhar 2,5 bilhões de dólares em uma única jogada: apostar todas as fichas na Cerebras, startup que promete revolucionar a inteligência artificial ao mover o processamento do mundo virtual para o físico. Não é só mais um aporte mirabolante no Vale do Silício — é o sinal mais claro até agora de que o futuro da IA não estará apenas nos data centers, mas sim nas máquinas que interagem diretamente com o mundo real.

A decisão do Eclipse chega em um momento crucial para o Brasil e os países de língua portuguesa, onde a corrida pela inovação tecnológica ganha força, mas ainda enfrenta desafios como infraestrutura limitada e mão de obra especializada. A Cerebras, conhecida por seus chips gigantes que processam dados em velocidade recorde, representa uma virada de chave: ao trazer a IA para o “mundo físico”, ela pode impulsionar desde fábricas automatizadas até sistemas de saúde mais precisos. Para nós, isso significa que, em vez de apenas consumidores de tecnologia estrangeira, podemos nos tornar protagonistas em áreas como agricultura de precisão ou monitoramento ambiental, setores onde o Brasil já tem expertise, mas precisa de ferramentas mais avançadas.

Se a aposta do Eclipse der certo, não será apenas mais um sucesso do Vale do Silício — será o prenúncio de uma nova era em que a IA deixará de ser uma promessa abstrata para se tornar uma força tangível, moldando desde as ruas das nossas cidades até os campos das nossas fazendas. O próximo passo? Torcer para que, desta vez, o Brasil não fique de fora da revolução.

🇪🇸 Resumen en Español

La apuesta millonaria de Eclipse por Cerebras marca un antes y después en la carrera por dominar la inteligencia artificial física, un campo donde los modelos tradicionales chocan con las limitaciones de la computación clásica. La firma de capital riesgo, con su inversión récord de 2.500 millones de dólares, no solo valida el potencial de los sistemas de IA integrados en hardware especializado, sino que redefine el terreno de juego para startups emergentes.

Este movimiento refleja un cambio de paradigma: la IA ya no se limita a procesar datos en la nube, sino que se desplaza hacia aplicaciones tangibles, desde robótica hasta sistemas autónomos. Para el público hispanohablante, la noticia subraya una oportunidad clave: la región, con su creciente ecosistema tecnológico y mano de obra especializada, podría posicionarse como polo de innovación en este sector. La pregunta ahora es si los inversores locales y las empresas adoptarán esta visión o quedarán relegados en una industria cada vez más dominada por gigantes extranjeros.