Key Takeaways

  • Invetta raised $311 million to build a chemical sequencer that maps life's chemical code rather than its genetic code.
  • Viswa Colluru states that 99 percent of the chemical matter inside humans, plants, and nature remains entirely unmapped by science.
  • The company avoids the software licensing trap, choosing to develop and own its own drug pipeline directly.
  • Model hallucinations do not hold back AI biology; the real bottleneck is physical wet-lab synthesis and the energetics of assembling molecules.
  • Invetta focuses on massive chronic markets, including oral treatments for asthma and eczema, alongside weight maintenance after GLP-1 drugs.

The True Frontier Is Chemical Code, Not Genetics

Genomics mapped the blueprint, but biology actually runs on small molecules. Most biotech startups spend billions analyzing DNA sequences that offer little direct control over cellular function. Invetta raised $311 million to attack the unmapped chemistry that sits downstream of genetics.

Colluru describes the mission in direct terms: “we're building the easiest way to describe it is think of it as a sequencer, but for life's chemical code instead of life's genetic code.” The gap in scientific knowledge is wider than most founders realize. As Colluru notes, “99% of what makes up you, a tomato in your garden or a random sample in the Amazon rainforest is still a mystery to science.” Mapping those molecules unlocks thousands of unstudied mechanisms for treating disease.

The Platform Trap: Why Software Models Fail Alone

Many tech founders enter biology thinking they can build an algorithm, license predictions to pharmaceutical giants, and collect pure-margin software royalties. Colluru rejects this playbook. Selling platform access creates low-margin consulting shops, while the real enterprise value in biotech has always come from owning the underlying therapeutic assets.

Colluru is blunt about how value accrues in the life sciences: “throughout the history of the industry, there's only been one way to build a big company in this space, and that is to make and own drugs.” He doubles down on this rule: “If you want to build a large and impactful drug company the rule is super simple. Make drugs and make drugs that matter.” By owning the intellectual property for actual compounds, a company captures the upside of clinical success instead of settling for SaaS fees.

Atoms Beat Tokens: The Wet-Lab Reality

Silicon Valley tends to view drug design as an information problem that larger transformer models will solve automatically. In reality, generating chemical structures on a GPU is easy; manufacturing them in physical space is brutal.

Colluru points out that digital predictions constantly crash into real-world constraints: “99% or more of the chemistry that any particular model comes up with is constrained by the physics and energetics of putting it together.” When an algorithm invents a theoretically perfect binder, human scientists still have to synthesize it step by step in a physical wet lab. The speed limit of drug discovery is chemistry and validation, not token generation.

Instead of chasing crowded oncology targets where marginal gains are small, Invetta applies this engine to massive everyday conditions. They target non-steroidal oral pills for asthma and eczema, plus long-term weight maintenance therapies for patients finishing GLP-1 cycles. These indications affect hundreds of millions of people who need accessible daily therapies rather than hospital infusions.

What to Do With This

Audit your core technology roadmap this week. If you are building an AI platform that outputs designs, recommendations, or simulations for a physical industry, identify the exact physical bottleneck your customers face when executing those designs. Shift your product engineering budget away from marginal model improvements and toward automating the physical synthesis, manufacturing, or testing loop that validates your output.