Simulacra AI: Foundational Model for Quantum Chemistry

Simulacra AI has successfully completed the training of its foundational model for quantum chemistry. 

This innovative Large Wavefunction Model supersedes traditional brute-force techniques by predicting the complete electronic wavefunction of a molecule. This provides the highest standard of accuracy, eliminating the need for exponentially increasing computational resources.

That flips the unit economics of simulation: ~29x cheaper than Orbformer today and 100x–1,000x cheaper beyond 30 atoms, running on commodity GPUs.

This unlocks creations of new drugs, high-energy-density batteries, advanced semiconductors, light materials for humanoid robots and aerospace industry.This capability enables the development of new drugs, high-energy-density batteries, advanced semiconductors, and lightweight materials crucial for the aerospace industry and humanoid robots.

GTM is zero-friction – Simulacra sells high-fidelity quantum data into existing in-silico pipelines – so pharma and materials teams get immediate lift in hit rates and time-to-clinic. 

It will create literally an avalanche effect in the drug discovery space and here is why:

• Modern drug discovery follows this pipeline: Molecular Candidate Generation → Simulation / Filtering → Preclinical Trials → Clinical Trials.

• The majority of these drugs are now trapped by the Simulation / Filtering bottleneck as existing methods (Hartree-Fock / DFT / CCSD(T)) are either inaccurate or can’t scale beyond simple molecules.

• Roughly 70% of all drug failures come from flaws in the Simulation / Filtering stage, due to inaccuracy of current methods.

• Simulacra AI foundational models enable precise molecular data generation, removing the bottleneck and unlocking billions of dollars trapped in the initial stages of the pipeline.

These drug discovery pipelines receive substantial funding. Approximately 50% of all venture capital investments, totaling $5.6 billion, have been directed toward the initial stage – AI-based Molecular Candidates Generation. However, Simulacra is poised to advance aggressively beyond this initial step.

Simulacra AI

We're excited to announce that our portfolio company Simulacra AI completed pre-training of the most accurate foundational model in quantum chemistry. Large Wavefunction Model replaces classical quantum chemistry simulation with:

• 29× lower cost than Microsoft’s Orbformer.

• 100×–1,000× cost advantage beyond 30+ atoms.

• Quantum chemical accuracy without exponential scaling.

After 30 atoms, everyone else breaks, whereas Simulacra AI’s LWM keeps scaling, which makes it the only viable option for molecular simulations that matter in practice.

Venture capital has heavily invested in AI-driven pipelines for materials and drug discovery in recent years. This investment has created substantial market demand for more accurate and affordable molecular simulation data — a need that remains unfulfilled.

So far in 2024, biotech and health companies have pulled in around $5.6 billion across 110 Series A rounds, per Crunchbase data. That accounts for 53% of all funding at the Series A stage, which is a closely watched barometer for the startup ecosystem. – Crunchbase

Currently every AI drug and materials discovery startup is racing into the same wall:

They can generate infinite molecular candidates. They cannot simulate them accurately at scale.

Moore’s Wall in Drug Discovery

Main goal of drug discovery is to find a molecule that cures disease or exhibits other important properties (binding affinity, toxicity, …) and here is how’s it currently done:

• First we generate 1 billion molecular candidates (costs are ~$100K) using AI. This is where the majority of companies are spending their resources (AlphaFold, RosettaFold, …)

• The majority (up to 99.999999%+) of these candidates are bad (hallucinations, physically impossible, …), but some are good and our primary objective is to identify good ones.

• The Simulation/Filtration phase is a crucial step where tens of thousands of promising candidates are identified. This process is costly (>$10M per run), and represents a significant bottleneck in terms of both computational resources and accuracy.

• Then hundreds of candidates go to pre clinical trials and single candidates are ready for clinical trials. The cost of error at the clinical trial stage can balloon up to $1.8B if the Simulation/Filtration stage was done poorly.

Simulation today forces an impossible choice between “cheap and useless” and “expensive and useless” methods:

• Cheap + Inaccurate = Useless (Hartree–Fock / DFT) – fails in non-convex chemistry, strongly correlated bonding, long-range charge transfer, delicate covalent interactions, open-shell and multi-reference transition-metal complexes, mispredicts stability… The list is endless. 

• Intractable + Accurate = Useless (CCSD(T)) – “gold-standard”, too slow, too expensive, only possible for smaller molecules and therefore useless.

This is why 70% of all drug failures trace back to faulty Simulation/Filtration predictions – not biology, not clinical strategy – bad physics simulation.

Without accurate simulation, every AI pipeline stalls. Simulacra AI is the enabling layer the entire sector depends on.

Large Wavefunction Models

Simulacra AI introduced the first of a kind Large Wavefunction Model that eliminates the trade-off. This foundation model that directly predicts the wavefunction, not just properties derived from it.

Simulacra AI’s pre-trained model makes quantum-accurate simulation computationally trivial, enabling systems far larger than any classical methods:

• Accurate enough to replace CCSD(T) – “gold standard”

• Fast enough to run inside of active R&D pipelines

• Cheap enough to scale to hundreds of thousands of molecules

• 29× lower cost than Microsoft’s Orbformer

• 100×–1,000× cost advantage beyond 30+ atoms

• Runs on commodity GPUs

Conviction 

Simulacra AI enables quantum chemical accuracy without exponential scaling – this completely flips the unit economics of molecular simulation and more importantly it increases humanity's chances to develop drugs faster and with higher success rate.

This unlocks breakthroughs previously blocked by compute:

• High-energy-density batteries

• Advanced semiconductors

• Aerospace and recyclable materials

• New therapeutics and GLP-1 successors

Simulacra AI sells high-fidelity datasets into existing in-silico pipelines, which means that no workflow change is required from the pharmaceutical companies and guarantees the fastest possible adoption rate.

Our conviction in Simulacra AI is clear: they have solved the core technical problem that stalled an entire industry. The science is validated. The market is here now. The inflection point is commercialization.

Simulacra AI is not competing with AI drug discovery companies – They are the layer that determines which of them succeed.

Cyber.Fund is committed to funding foundational, category-creating companies on the verge of scale – and Simulacra AI embodies this singular opportunity.

More details are available here: https://arxiv.org/abs/2511.07433