A New AI Lab has officially burst onto the scene. It secured the biggest seed round in artificial intelligence history. Mirendil, founded by ex-Anthropic researchers, recently closed a massive $200 million funding round. The startup achieved an impressive post-money valuation of $1 billion. This historic milestone happened before the company shipped a single product. They have not even published a scientific paper.
This massive valuation without a tangible product might seem unusual. However, it represents the new normal for frontier AI development. Investors are no longer just buying current revenue. Instead, they are backing raw scientific credibility. This New AI Lab is betting on elite researchers to build self-accelerating software.
Who Built the New AI Lab and Why It Matters
The founding team at Mirendil represents a deep reservoir of elite talent. Co-founders Behnam Neyshabur and Harsh Mehta met at Google in 2019. They later moved to Anthropic in late 2024. Neyshabur, who served as a core reasoning researcher at Gemini, acts as CEO. CTO Harsh Mehta previously built Anthropic’s in-house auto-research platform.
The team also includes Shayan Salehian, an early ML engineer from xAI. Tara Rezaei, a brilliant 23-year-old MIT graduate and Olympiad medalist, completes the founding quad. Together, they have assembled a team of 20 elite researchers. These scientists hail from Anthropic, OpenAI, Google DeepMind, and xAI. These are seasoned specialists who understand the core mechanics of contemporary AI.
Mirendil wants to build AI systems that can automate AI research itself. Rather than just answering questions, the system acquires deep expertise over time. It can design experiments, manage GPU compute, interpret results, and debug failures autonomously.
Typically, organizations require a comprehensive Ai Consulting Strategy before deploying models. Mirendil wants to simplify this process. They train frontier models that are experts at AI research and development. The goal is to allow organizations without massive machine-learning teams to build highly specialized tools.
How the New AI Lab Fits Into the Funding Frenzy
This New AI Lab is part of a broader wave of “neo-labs” raising extraordinary capital. These startups are led by proven leaders who secure funding before building commercial software. For instance, former OpenAI Chief Scientist Ilya Sutskever raised $1 billion for Safe Superintelligence. That company achieved a $5 billion pre-money valuation. Similarly, former OpenAI CTO Mira Murati has been raising $2 billion for her venture, Thinking Machines Lab.
When selecting a Top Ai Development Company, enterprises usually look for established products. In the frontier AI sector, however, investors fund aspirations rather than business metrics. Venture capitalists are supporting technical visions rather than immediate balance sheets.
If you are exploring the technical mechanics of AI, understanding What Is Application Software is a fundamental first step. Yet, Mirendil operates at a much deeper level of the stack. They are building the infrastructure that will generate next-generation software autonomously.
Why Nvidia Keeps Showing Up in Seed Rounds
Nvidia’s participation in Mirendil’s seed round reveals a calculated strategic move. The semiconductor giant continues to invest in emerging AI startups. The logic behind this approach is clear. Frontier model developers face severe compute shortages during training.
By backing elite research teams early, Nvidia accomplishes two goals. First, it secures future demand for its advanced hardware. Second, it gains an early preview of how the technical frontier is evolving. This allows Nvidia to build a robust sales pipeline while gathering valuable technical intelligence.
As startups scale, they often adopt a Rag System Architecture Design to optimize how data is processed. Nvidia’s early involvement ensures that future architectures are optimized for their proprietary hardware. It is a mutually beneficial arrangement for both parties. It secures Nvidia’s dominance at the hardware layer.
OpenAI’s Broadcom-Built Inference Chip “Jalapeño”
While Mirendil automates the software side, OpenAI is tackling the hardware bottleneck. The company recently unveiled “Jalapeño,” its first custom AI processor. Developed in collaboration with Broadcom, this chip is designed specifically for large language model inference.
This move is a direct response to broken unit economics. In FY2025, OpenAI generated $13.07 billion in revenue. However, they suffered an operating loss of $20.92 billion. Running inference entirely on general-purpose Nvidia GPUs proved to be financially unsustainable. Jalapeño strips out unnecessary silicon overhead to serve models more efficiently.
The chip reached tape-out in a rapid nine-month development cycle. This was achieved because OpenAI used its own AI models during the chip design process. This development showcases the power of AI-assisted engineering. It is a trend that aligns perfectly with the list of top Ai Agents Companies 2025.
To find out more about how autonomous search engines are tracking these advancements, you can check out the Stripe Directory Ai Agents Search Engine. OpenAI expects to start deploying Jalapeño in late 2026. They target full-scale gigawatt deployments with Microsoft by 2028. This custom processor marks a major shift toward vertical integration in the AI industry.
Anthropic Accuses Alibaba of “Claude” Distillation Attack
While hardware and software advance rapidly, AI security faces major challenges. Anthropic recently accused Chinese e-commerce giant Alibaba of launching a massive distillation attack against Claude. According to a letter sent to the US Senate Banking Committee, Alibaba created 25,000 fraudulent accounts.
These fake accounts generated over 28.8 million exchanges with the model. The campaign ran between April 22 and June 5, 2026. The attack targeted some of Claude’s most valuable capabilities. These include agentic reasoning, software engineering, and long-horizon tasks. Following the news, US-listed Alibaba stock fell roughly 2.7% to a 52-week low.
This is known as a distillation attack. In this setup, a competitor trains a “student” model on the outputs of a “teacher” model. This technique replicates a proprietary system’s performance at a fraction of the cost. For enterprises, protecting model outputs is becoming as critical as Claude Fable 5 Smart Contract Security.
As AI tools become integrated into business logic, firms must evaluate how they handle sensitive workflows. This applies to everything from Ai In Customer Support to complex backend integrations. Anthropic is now urging Congress to close loopholes. According to a report by Reuters, they want to prevent foreign entities from bypassing security guardrails.
Expanding AI into Enterprise and Creative Fields
These massive technological breakthroughs are rapidly reshaping the global job market. Many technical professionals are looking at the Top 10 High Paying Remote Jobs 2026 2024 to see where the industry is heading. Engineering and data science roles remain highly lucrative as companies rush to build proprietary AI pipelines.
At the same time, organizations are seeking an Ai Development Company In Berlin and other tech hubs to localize their operations. AI-driven automation is also streamlining departments like human resources. Specialized tools for Ai In Hr are now managing recruiting, onboarding, and talent retention.
We are also seeing AI enhance customer experiences. For example, implementing Ai In E Commerce Customer Experience solutions helps brands provide personalized recommendations and seamless support. This goes hand-in-hand with standardizing Robotic Process Automation across administrative pipelines to eliminate repetitive manual labor.
Meanwhile, other organizations are fostering open-source development. The Sentient Foundation recently announced a $42 million grant and research program. These funds will support developers, researchers, and companies building open-source AGI tools. This ensures that frontier AI remains accessible to a broader community.
Conclusion
The launch of Mirendil proves that the frontier AI ecosystem is moving at breakneck speed. From self-improving software to custom silicon like OpenAI’s Jalapeño, the boundaries of technology are constantly expanding. As AI labs continue to raise historic rounds, the focus is shifting from simple chatbot platforms to autonomous research engines. Security challenges, such as the distillation attacks alleged by Anthropic, highlight the high stakes of this global race. Only time will tell if these astronomical valuations will yield the scientific breakthroughs investors are betting on.