Zk Proofs: The Key to DePINs’ Edge in AI’s $7.6 Trillion Future?

by

Experts Say Zk Proofs Give DePINs an Edge as AI Trust Demands Rise

Goldman Sachs predicts companies will invest $7.6 trillion in artificial intelligence (AI), but this estimate depends on how long current AI computer chips remain effective. While decentralized networks could significantly lower costs, they currently struggle with speed. Experts believe these networks will only succeed long-term if they focus on ensuring data accuracy and trustworthiness, even if it means sacrificing some speed.

  • Key Takeaways:

  • Goldman Sachs cites a $7.6 trillion spend by 2031, depending on whether chips last more than 3 years.
  • StealthEX and Cysic experts warn that DePIN latency limits decentralized AI to batch jobs over live chat.
  • Onchain firms like Maple may bridge the $5M to $50M credit gap for Tier 2 data centers by 2028.

The $7.6 Trillion Baseline

A recent Goldman Sachs report shifts the debate from whether artificial intelligence (AI) demand exists to which supply-side factors will determine the actual cost of the build-out. The report projects $7.6 trillion in AI capital expenditure as a baseline but emphasizes that this figure is highly sensitive to “swing variables,” including the useful life of AI silicon.

How long chips remain useful is a key concern. Fast-paced advancements could make standard chips outdated in just three years, even though they usually last four to six, leading to much higher expenses. However, a system where older chips are repurposed for less demanding jobs, like processing data, could help keep costs stable.

How much money companies invest in AI infrastructure over the next five years will likely depend on how complicated data centers are and how quickly computing needs change. Building out this infrastructure is also taking longer than expected due to limited power availability, a shortage of skilled workers, and difficulties getting the necessary electrical equipment.

A separate report, meanwhile, frames this staggering infrastructure expenditure as the cornerstone of an emerging “machine economy.” In this paradigm, AI agents become the primary economic actors, executing high-frequency transactions and managing resource allocation independently. The report’s authors contend that legacy financial systems, characterized by slow settlement cycles and rigid know your customer (KYC) frameworks, are fundamentally ill-equipped for the velocity of agentic commerce.

Decentralized Infrastructure and the Latency Trade-off

Consequently, it positions crypto and decentralized protocols as the essential, permissionless “economic rails” required to facilitate this shift. However, skeptics remain wary, questioning whether decentralized physical infrastructure networks (DePINs) can truly mitigate AI’s ballooning capital requirements.

Vadim Taszycki, head of growth at StealthEX, notes that while decentralized networks can offer significant cost savings, they face physical limitations. While a decentralized provider like Akash might rent an H100 GPU for $1.48 an hour compared to $12.30 on Amazon Web Services, the trade-off is speed.

As a crypto investor, I’ve been following the debate around centralized versus decentralized AI infrastructure. What I’m hearing is that big cloud providers have a real advantage right now when it comes to speed. They can run AI tasks incredibly quickly because all their powerful GPUs are physically located in the same building, connected by super-fast cables. Decentralized networks, where GPUs are spread out across the globe and connected via the internet, just can’t match that speed – we’re talking milliseconds of delay. That makes decentralized options good for things like processing large datasets or refining AI models, but not so great for applications needing instant responses, like a chatbot where even a tiny delay would ruin the experience.

Leo Fan, the founder of Cysic, agrees and believes running tasks in a decentralized way isn’t ideal when you need things to happen very quickly. However, Fan points out that measuring speed isn’t the best way to compare decentralized systems to large, established cloud providers like AWS.

According to Fan, the real challenge isn’t simply having enough computing power, but rather figuring out how to find resources, schedule tasks, and verify their authenticity. He believes the key to progress isn’t lowering the cost per token, but ensuring things can be reliably verified. He explained that technologies like secure enclaves and zero-knowledge proofs enable decentralized networks to excel in areas where trust and verification are more important than speed.

Onchain Credit and the Funding Gap

The conversation is moving beyond just the technology itself to how these large projects are financed. While traditional lenders have plenty of money, they often miss out on smaller or unique opportunities. Onchain lending offers benefits like letting everyday investors share in the profits from data centers, which were previously only available to big institutions. Platforms such as Maple and Centrifuge can also pool loans ranging from $5 million to $50 million – a size that larger firms like Apollo often avoid because the cost of evaluating these loans is too high compared to the potential profit.

Onchain credit also allows for new “pay-per-use” models for computing power, meaning income changes based on how much a GPU is used. These models are a better fit for sharing revenue through tokens than long-term, fixed leases.

Despite this potential, experts identify four “gates” that remain closed to institutional adoption: legal enforceability in bankruptcy courts, the lack of tamper-evident oracle infrastructure for servicing covenants, regulatory uncertainty for billion-dollar tranches, and unstandardized tax and accounting products.

Most experts believe it will take 12 to 24 months before we see significant on-chain activity for moderately sized syndicated loans. Fully on-chain mezzanine debt – a more complex financial product – is likely still three to five years away. We’ll probably see the first successful examples from newer companies, rather than established leaders like Coreweave.

Read More

2026-05-14 08:27