Fabric
Without privacy or trust we cannot have a functioning civil society. Yet, those two ideals seem to be at odds with one another. If all information is transparent we have perfect trust but sacrifice privacy. The result is a society prone to mass manipulation, psy ops and suppression. If all information is private we have perfect privacy but can barely learn from one another or trust each other. FTX and Wirecard kept their books private. How can this dilemma be resolved? What if we could balance privacy and trust in more nuanced ways through technology?
Moon math to the rescue. Over the last few decades numerous cryptographic techniques have been developed to solve the conflict between privacy and trust. Zero Knowledge Proofs (ZKP) empower us to prove a fact without revealing its underlying information. Fully Homomorphic Encryption (FHE) enables us to run computations on encrypted data sets. With multi party computation (MPC) we can jointly compute a function over their inputs while keeping those inputs private (MPC).
The blockchain industry and militaries have been early pioneers of implementing those techniques at rather small scales. On a high level broader commercialisation requires (1) orders of magnitude higher performance in terms of proof generations, verifications and computations, (2) drastically lower costs per operation and (3) standards and tooling for seamless implementations. How do we get there?
Enter Fabric Cryptopgraphy, a recent addition to the Inflection portfolio which just announced its $33M Series A round. We co-led their $6M Seed round in 2023 and participated in their Series A.
In 2022 the Fabric founders came to the conclusion that cryptographic algorithms will need a native processing unit, just like AI has GPUs / TPUs or general purpose computers have CPUs. This realisation drove their development of verifiable processing units (VPU) - a general purpose processor for cryptography that comes with its own instruction set and software stack. An NVIDIA for cryptography that can accelerate ZKPs, FHE, MPC and many other classes of crypto 2.0 algorithms down the road.
Our Thesis
We couldn't be more excited about the problem set Fabric is going after. Our digital-physical fabric is prone to numerous attacks - spanning impersonation, data theft and deep fakes. Such problems are gigantic in terms of societal and economic impact for themselves. In the context of military conflicts decided by autonomous weaponry they become existential. Think of AI as offence and cryptography as defence. The rat race for programmable cryptography just started and will accelerate over the next decade.
What distinguishes Fabric from others is the design of their architecture. An FPGA centric approach would have come with the highest degree of programmability and flexibility but lower performance. A fixed function ASIC would have optimised for one specific algorithm only to yield maximum performance but compromising flexibility. Instead, Fabric develops a general purpose crypto chip that optimises for performance and flexibility with reasonable trade offs.
We consider this a dominant strategy as crypto software is ever evolving and hyper dynamic. Algorithms that work in a certain way today might operate very differently in three months from now. To borrow an analogy from biology: in predictable environments with fixed rules and marginal changes specialisation wins due to pattern recognition. In messy environments with changing rules the most adaptive, generalist organisms win. New cryptography certainly falls into the second category for the foreseeable future.
Their incredible team of world class semiconductor and cryptography talent is growing. If you consider a collaboration feel free to reach out or check out open positions here.