Andrew Feldman of Cerebras Systems Discusses the Future of AI Scaling at TechCrunch Disrupt 2026
Cerebras CEO Andrew Feldman discusses wafer-scale AI, CS-4, OpenAI deals, and how data centers and power now limit AI scaling.
Key Takeaways
- Cerebras Systems is challenging conventional AI hardware by scaling wafer-scale computing, offering large specialized processors for demanding AI workloads through on-premise systems and cloud platforms.
- CEO Andrew Feldman will discuss AI scaling limits, rising compute and energy demand, and alternative infrastructure approaches at TechCrunch Disrupt 2026.
- Cerebras has moved from a research bet to commercial AI infrastructure, highlighted by its $5.5 billion IPO, a multiyear OpenAI deal for 750 megawatts, and the launch of its CS-4 wafer-scale system.
- The key bottleneck in AI scaling is shifting from chip design to physical deployment, including data centers, power, cooling, manufacturing capacity, and large specialized facilities.
- Cerebras’ expansion shows that future AI progress depends not only on more powerful hardware, but also on the supply chain’s ability to deliver compute at scale.
Meta description: Cerebras Systems brings wafer-scale computing and CS-4 to TechCrunch Disrupt 2026 as OpenAI deals, data centers, and compute limits shape AI scaling.
Tags: Cerebras Systems, wafer-scale computing, AI infrastructure, CS-4, TechCrunch Disrupt 2026
Featured image: abstract silicon wafer and server-grid geometry, cool blue tones, no real people, no company logos, no text.
Background: A Different Bet on AI Hardware
AI models keep getting more capable, but each capability jump increases demand for compute, energy, and physical infrastructure. Cerebras Systems has spent the past decade challenging the assumption that more powerful AI must rely on conventional chip architectures. The company built its business around wafer-scale computing and today provides AI compute through on-premise systems and its cloud platform.
Andrew Feldman, Cerebras Systems CEO and co-founder, will take the Disrupt Stage at TechCrunch Disrupt 2026 for "Can AI Keep Scaling?" He will discuss the rising demand for compute, energy, and infrastructure, how Cerebras approaches those constraints differently, and what may follow if today's AI hardware reaches its limits.
Feldman's path into Cerebras runs through earlier infrastructure businesses. He co-founded Cerebras in 2015 after years of building companies around computing infrastructure. Before Cerebras, he co-founded and led SeaMicro, an energy-efficient microserver startup acquired by AMD in 2012. Earlier, he held leadership roles at Force10 Networks and Riverstone Networks.
Development: From a Hard Idea to Commercial AI Infrastructure
Cerebras began with a problem many considered impractical: bringing wafer-scale computing to market. Instead of cutting a silicon wafer into separate chips, the company developed a processor built on the wafer itself, with an architecture designed specifically for demanding AI workloads.
That approach is now scaling as AI compute demand grows. Cerebras raised $5.5 billion in its May IPO and signed a multiyear agreement with OpenAI to deploy 750 megawatts of Cerebras systems from 2026 through 2028. In August, the company introduced CS-4, the latest generation of its wafer-scale AI infrastructure.
The OpenAI deal and the IPO show that wafer-scale systems are no longer only a research bet. They are becoming part of the commercial supply chain for large AI workloads. The question is no longer only whether Cerebras can build a large processor, but whether the wider supply chain can deliver enough power, facilities, and production capacity to make that processor useful at scale.
Latest Progress: Power and Factories Become the Bottleneck
More powerful processors alone do not solve the scaling problem. Those systems need data centers, electricity, cooling, and manufacturing capacity. Cerebras is already confronting that challenge.
In August, the company reported more than 600 megawatts of data center capacity live or under contract for delivery by the end of 2027. It also said it was increasing manufacturing capacity more than tenfold during 2026. In addition, Cerebras plans to bring its first European data center capacity online this year and expand to 200 megawatts there by the end of 2027.
The numbers point to a broader shift in AI infrastructure: the limiting factor is moving from chip design to physical deployment. A new accelerator generation matters only if it can be installed in powered buildings, cooled reliably, and produced fast enough to meet demand. Cerebras' wafer-scale model makes that constraint visible because its systems are large, specialized, and dependent on facilities that can accommodate them.
Current Stage: What TechCrunch Disrupt 2026 Adds
Cerebras pursued wafer-scale computing long before the current AI infrastructure boom. Now it is scaling its computing and manufacturing capacity as demand accelerates. At Disrupt, Feldman will explore what growing demand for compute, energy, and infrastructure means for AI's future and what could happen if conventional hardware can no longer keep pace.
For founders, investors, and technology leaders working around AI infrastructure, the session puts the constraints into context: where compute demand is heading, what it takes to support it, and where today's hardware could hit its limits. TechCrunch Disrupt 2026 takes place October 13-15 at Moscone West in San Francisco, with 200+ sessions across six industry stages, roundtables, and breakouts. More than 10,000 founders, investors, operators, and tech leaders are expected, along with 250+ speakers and 300+ exhibiting startups. Beyond the agenda, matchmaking, dealmaking, and networking create opportunities to connect with the builders shaping what comes next. Attendees can bring a co-founder, colleague, or peer at 50% off.
Source image credits: TechCrunch; Eric Slomonson, The Photo Group.
Comments
No comments yet. Be the first to comment.