LooksRare

Pinata: From IPFS Storage Tool to AI Agent Infrastructure — Kyle Tut, Matt Ober, and the Evolution of Web3’s File Network

Pinata is no longer just an “NFT storage tool.” Its legal entity is Pinata Technologies, Inc., and its public website and terms show that it continues to operate out of Omaha, Nebraska. Historically, it started with IPFS pinning, NFT media distribution, and dedicated gateways, but by 2026 the homepage clearly presents two parallel tracks: IPFS file storage and AI agent hosting. The homepage literally says “AI agents or file storage—choose your starting point,” which shows that it has expanded from “Web3/NFT infrastructure” into a hybrid platform for both file infrastructure and agent infrastructure. Its documentation still emphasizes that it has been at the forefront of IPFS since 2018 and describes itself as “the biggest IPFS provider in the Web3 space.” Today, Pinata’s real-world position is closer to a developer infrastructure company than a narrative-first Web3 brand. The homepage says it is trusted by “600,000 developers.” Its product surface now spans IPFS uploads, dedicated gateways, private IPFS, Groups, KV Storage, Workspaces, the Files API, OpenClaw/Hermes agents, a CLI, and an agent template marketplace. That breadth shows it is not selling one isolated feature, but a full managed stack so developers do not have to maintain the underlying storage and runtime themselves. There is not one founder but two. Public funding coverage, company history posts, and company/job materials all point to Kyle Tut and Matt Ober as the founders. Kyle appears as Kyle Tut in most public materials, but some company/job sources use the fuller name Kyle Tautenhan. The division of roles is unusually clear: Kyle is the CEO focused more on market framing, product narrative, fundraising, BD, and community-building; Matt is the CTO focused more on architecture, developer experience, product simplification, and technical implementation. If Pinata must be summarized in one sentence, it solves a long-term infrastructure problem: on-chain storage is expensive, but off-chain data still has to remain stable and retrievable. In the Greylock interview, Kyle explicitly recalls realizing around 2018 that storing large data directly on chain was prohibitively expensive, so real applications would inevitably push data to IPFS. The trouble was that early IPFS was not fast or stable enough for production. Their company goal became simple: make IPFS as fast and as stable as possible. TechCrunch described Pinata as an NFT-focused media distribution/infrastructure platform; by 2026, that same file capability had extended into agent infrastructure. On family background, the only honest phrasing is: public information is limited / cannot currently be confirmed. Across Pinata’s official pages, founder author pages, the Greylock interview, and publicly indexed professional profiles, there is almost no reliable disclosure of birth dates, birthplaces, parents’ occupations, family class position, or childhood resources for either Kyle Tut or Matt Ober. What can be confirmed is that both are strongly tied to the Omaha, Nebraska startup and developer ecosystem, but that does not justify speculation about their original family background. Kyle’s education is clearer, but it is not a “prestige school myth” story. Public professional materials show that he attended the University of Nebraska at Omaha, with public profile summaries indicating marketing and management-related studies. More important than the degree name is the entrepreneurial environment around UNO. A 2025 Omaha Magazine interview directly linked Kyle’s development to the university’s entrepreneurship ecosystem, and Kyle himself said he had always wanted to start his own company, describing UNO’s entrepreneurship program as giving him the basics needed to do that. This matters because it explains how a non-technical founder eventually entered a highly technical blockchain infrastructure field. Matt’s education is more traditionally engineering-oriented. Public profiles link Matt Ober to the University of Nebraska-Lincoln, and publicly indexed bios describe his academic background as electrical and electronics engineering. In practical terms, the two founders had a naturally complementary structure from the beginning: Kyle leaned toward market, organization, and entrepreneurship, while Matt leaned toward engineering and systems implementation. That is exactly the type of pairing that fits a developer tools company whose product must be deeply technical but usable by people who are not protocol-level specialists. Kyle’s first truly representative work experience was not in crypto at all, but in a motorsports startup. In the Greylock interview, he says that his first job out of college was at a motorsports startup working around audio communications for professional race teams. Public LinkedIn snippets add that he handled brand awareness and operations-related work for RaceNote, including Google AdWords, Facebook Ads, social media marketing, and related responsibilities. The significance is not the racing industry itself; it is that this stage trained him in process, operations, brand-building, and selling products to a specific vertical. Kyle entered crypto not because he wanted to become a trader, but because he wanted to build a company. In the Greylock interview, he recalls buying Bitcoin in 2015 and becoming genuinely interested in Ethereum and smart contracts in 2016. He explicitly says he was not a “crypto finance trader guy,” but someone who had always wanted to build his own company. In April 2017, he quit his existing job and jumped into blockchain and crypto full time without already having a fixed employer or fully defined destination. That decision matters because it moved him from being a local startup operator into a global hackathon and builder network. Kyle’s main way of learning was not formal technical retraining but full immersion in hackathons. Because he was based in Omaha, which was not a core crypto hub, he flew to New York for the 2017 Consensus Hackathon, then spent the summer flying to blockchain hackathons, building both technical understanding and a network of engineers. At the same time, he started a blockchain meetup in Omaha that eventually grew to around 500 people. It was through that local community that he met Matt Ober. This shows that Kyle’s growth path was not “technical founder turned business founder”; it was “non-technical founder forcing himself into technical environments until he could operate there.” Matt’s path into Pinata also began from an engineering pain point rather than NFT collecting. The company’s five-year history post states repeatedly that in 2018, while the two were building side projects and attending hackathons, Matt tried to use IPFS in real projects and had to spin up his own IPFS nodes. He found the process tedious and challenging, and that frustration became the immediate trigger for Pinata’s founding. In other words, Pinata did not begin with a grand abstract thesis and then search for a use case; it began with a developer problem the founders had already experienced firsthand. Before Pinata, the two had already gone through an earlier blockchain consulting phase. In the Greylock interview, Kyle says that before Pinata they had already started a consulting company building blockchain applications for different people. Publicly indexed professional and conference materials identify that earlier venture as BlockEra. The key point of that period is straightforward: by building blockchain applications for others, they kept running into the same recurring issue—data was too expensive to store on chain, and off-chain content infrastructure was still brittle. Pinata was essentially the generalized infrastructure problem extracted from that consulting experience. Pinata’s real origin point is ETHBerlin 2018, and it began by winning a hackathon. The official company history, PRNewswire, and multiple investor/media materials all align on this. Kyle and Matt arrived at ETHBerlin 2018 with two laptops and a simple idea, won the hackathon, and formed the earliest version of Pinata there. That mattered not merely because they won a competition, but because it validated that the IPFS infrastructure problem could be turned into an independent company. But winning the hackathon did not immediately create a smooth growth curve. Kyle says they initially thought they would raise money quickly and scale fast, but reality was the opposite. From 2018 to 2020, they struggled for close to two years. The reasons were twofold: NFTs were not yet a major market, and the broader crypto sector was in a crypto winter. In the Greylock conversation, he makes the point that being right too early often feels indistinguishable from being wrong. The way they survived the cold-start period was not financial engineering but extreme closeness to developers. Kyle says they kept attending hackathons in 2018, 2019, and 2020, kept calling projects, kept talking to builders, and kept publishing technical blog content that taught people how to build in the ecosystem. The company’s five-year reflection also emphasizes hackathons, in-person conversations, and internal dogfooding as ways to understand what developers actually needed. That means Pinata’s growth engine was not primarily token speculation or hype cycles; it resembled a classic devtools growth path built on docs, tutorials, community, case studies, and low-friction integration. The period from 2021 to 2022 was when Pinata received both capital validation and market validation. TechCrunch and PRNewswire both state that the company publicly disclosed a total of $21.5 million in funding, composed of a $3.5 million 2021 seed round and an $18 million 2022 Series A. Greylock and Pantera co-led the Series A; Greylock and Offline Ventures co-led the seed; other investors included Volt Capital, OpenSea, and Alchemy. That investor mix is meaningful because it spans mainstream VC, top crypto capital, infrastructure players, and application ecosystem participants. The real value of that financing was not just cash, but the visibility of its resource network. Greylock publicly described working with Pinata since 2021 and framed the company around making tokenized media work at scale. PRNewswire listed customers and partners such as OpenSea, DraftKings, Protocol Labs, Yuga Labs, Autograph, and LooksRare. That means Pinata was never just a local Midwest project; it had already become embedded across NFT marketplaces, crypto media infrastructure, developer protocols, and branded issuance ecosystems. Pinata’s earliest core “real asset” was not abstract brand equity but managed infrastructure. By 2022 the company was emphasizing 200+ server locations, a built-in CDN, Dedicated Gateways, Submarining, and pinning services. By 2024–2026, the core asset base had evolved into its gateway network, upload and retrieval APIs, unified SDK, account and access systems, workspaces, payment integrations, and managed agent runtime infrastructure—plus the customer trust that comes from case studies and repeated performance claims. Alongside those real assets, Pinata also cultivated a very deliberate layer of “influence assets.” The clearest example is Pinnie, the brand mascot. The company’s five-year post makes it explicit that Kyle and Matt did not want a cold, minimal, dark, “hackerman” aesthetic. They intentionally moved in the opposite direction and built a colorful, memorable character suited for hackathon circulation, drawing inspiration from CryptoKitties and the old multi-color Apple logo. Pinnie is not a cash-flow asset, but it is a memory asset, a community asset, and a brand-recognition asset. Pinata’s business model has gone through several layers of abstraction. At first it sold IPFS pinning and NFT media distribution. In 2022 its public narrative still revolved around being the “home for NFT media” and an NFT distribution platform. By 2024 it had begun to reframe itself as “The Internet’s Files API,” meaning that the upload, distribution, access control, CDN, and gateway capabilities forged during the NFT boom were being repackaged for broader developers and businesses. By 2026, the homepage and documentation had pushed that managed infrastructure logic even further into AI agents. At root, the company keeps selling the same promise: you do not need to build, manage, secure, and iterate the lower-level environment yourself. Its current public pricing makes its SaaS/infrastructure model very explicit. As of 2026, the pricing page shows four tiers: Free, Picnic, Fiesta, and Enterprise. The core billing units are not just storage anymore, but storage plus gateways, CDN, and agents. Free includes 1GB storage and 1 gateway; Picnic is $20/month with 1 agent, 1TB storage, and 1 gateway + CDN; Fiesta is $100/month with 3 agents, 5TB storage, and 3 gateways + CDN; Enterprise is custom. In other words, Pinata is not monetizing primarily through token issuance or vague brand partnerships—it is charging recurring subscription fees for infrastructure. It is also trying to turn content monetization itself into platform infrastructure. In 2025, Pinata introduced x402-based paid access, allowing users to attach payment instructions to private content so that payments flow directly to their own wallets. The documentation states this plainly: users can monetize private content, set custom prices, and receive USDC directly. At the same time, Pinata launched an MPP server that allows machines without a Pinata account to upload and download from IPFS through payment. That shows Pinata no longer wants to be only a storage layer; it is expanding into a content access and machine-payments layer. One of the most notable things about Pinata’s product history is that it has repeatedly rebuilt instead of clinging to legacy decisions. In 2023, Pinata’s official help center explicitly announced the sunset of Submarine, Private Files, and the Private API, telling users to migrate or export before December 1, 2023. But in 2024 it launched a new Files API and private file support, in 2025 it introduced Private IPFS, and in 2025 it openly admitted that its old API had become too complex and confusing as new features were added—hence the rewrite into V3 and a unified SDK. This shows two things at once: first, Pinata is not path-dependent and is willing to kill old implementations; second, it has imposed real migration costs on users when it changes course. The biggest strategic change in 2026 is that Pinata is now applying its managed infrastructure philosophy to AI agents. Official docs define Pinata Agents as hosted AI agents inside sandboxed containers, equipped with workspaces, terminals, connectors, and persistent state. In May 2026, the product expanded from initial OpenClaw support to Hermes support as well. The CLI then extends this by giving developers a faster loop for creating, inspecting, restarting, and snapshotting agents. Put differently, Pinata’s new story is not a rejection of its old capabilities; it is a reuse of them. Storage, gateways, permissions, state, and managed deployment are now being repurposed as agent infrastructure. Its 2026 partner map shows that Pinata is trying to move into the chain-execution and agent-workflow layer. Public blog posts show MoonPay agent templates for portfolio management, cross-chain execution, and fiat on-ramping. Around the same time, Pinata added Hermes support and joined the Legal Context Protocol effort launched with the AAA, Integra Ledger, Google, IBM, Circle, Wayfair, and others. Kyle Tut’s public quote for that initiative explicitly ties Pinata’s IPFS and content-addressing capabilities to making agentic commerce more trustworthy. That suggests Pinata is trying to extend its IPFS/data-verification role into machine commerce and agent transaction governance. The first major career-defining decision by the founders was to bet on off-chain content infrastructure in 2017–2018 instead of chasing hotter narratives. At the time, it was easier to chase token launches, DeFi excitement, or speculative hype. They instead focused on a dirtier, slower, more foundational problem: how large files are stored, retrieved, served quickly, and kept stable. The significance of that decision is that it let Pinata avoid many short-lived hype businesses and put it in a strong position once NFTs exploded. The second key decision was Kyle’s choice, as a non-technical founder, to enter technical communities directly instead of sitting at the edge of the industry doing content or light branding. That choice shaped the company’s later product character. Even though the CEO was not trained as an engineer, the company’s origin came from hackathons, developer communities, and technical documentation—not from celebrity access, influencer-driven NFT launches, or media arbitrage. That is why Pinata’s eventual growth logic feels more like devtools than like a traffic-driven NFT brand. The third key decision was the willingness from 2023–2025 to redraw product boundaries and rebuild APIs. This created friction in the short run, but the longer-term effect is clear. Pinata moved away from a patchwork of years-old APIs into a unified V3 API, then broadened from a Web3-specific tool into a more general Files API and Private IPFS system, and then extended again into agent hosting. Many infrastructure companies die because they keep piling features onto old systems they are afraid to replace. On its public product track, Pinata did not take that conservative route. The most important result Pinata achieved is not the dollar amount of its financing, but its position inside NFT and IPFS infrastructure. Its documentation calls it the biggest IPFS provider in Web3. A 2025 company analysis of the top 1,000 NFT collections said that 327 were on IPFS and that 288 of those were on Pinata. A 2024 OpenSea case study reported that Pinata accelerated metadata changes by 99.2%, brought near-zero downtime, and shortened development cycles by 65%. Since many of these numbers come from company or case-study framing, the safest conclusion is this: Pinata clearly held a very high-ranking position in the top tier of NFT media infrastructure. The main public controversies around Pinata are not about founder misconduct, but about product and infrastructure tension. There are three main categories. First is the tension of “managed decentralization”: Pinata helps make IPFS usable, but it is also a centralized service provider, so it is inevitably evaluated through the question of whether it is enabling decentralization or re-intermediating it. Matt has even published under the title “Everything Is Centralized Somewhere,” which shows the company does not avoid the issue. Second is the cost of sunsetting and rebuilding products, especially the 2023 shutdown of Submarine, Private Files, and the Private API. Third is operational reliability, most notably the June 12, 2025 outage. The June 2025 outage is the hardest public incident on Pinata’s record. The company’s own postmortem says that from 1:04 PM CDT to 3:26 PM CDT on June 12, 2025, uploads, file retrieval, and the public IPFS gateway were partially or fully unavailable. Pinata’s explanation was that a large Google Cloud outage propagated indirectly through Cloudflare dependencies, especially Cloudflare KV, which Pinata was using for gateway performance and metadata handling. In the postmortem, the company explicitly says “we failed at that” and committed to auditing lower-level dependencies. For infrastructure companies, the deeper issue here is not just the downtime itself, but the reminder that “multi-cloud” does not automatically eliminate hidden single points of failure. A longer-running criticism comes from the structural reality of NFT metadata and infrastructure itself. Pinata has long argued that NFT metadata should not remain on fragile centralized servers, and that point is valid. But the industry also therefore comes to rely on managed providers like Pinata as critical intermediaries. Academic research has likewise shown that a significant share of NFT metadata still sits on centralized platforms, carrying meaningful availability and censorship risk. Pinata’s business value was born from that industry flaw; but by the same logic, as long as the ecosystem depends on managed infrastructure, it can never fully occupy the position of absolute decentralization purity. This is not a claim that Pinata is uniquely at fault—it is a structural criticism of the layer it occupies. If you look at Pinata’s status today, it has already gone through three identity changes. First, it was a small team solving IPFS usability for on-chain builders. Second, it became a major infrastructure provider for NFT media distribution and dedicated gateways. Third, by 2026, it is actively rewriting itself as a platform for autonomous file storage plus agent infrastructure. Whether that third identity ultimately becomes as successful as the second is still too early to say, but judging from the homepage, product docs, pricing, templates, and participation in external protocols, this is not a mere experiment—it is an explicit strategic redirection. Placed accurately in the market, Pinata and its founders are best understood as intermediate infrastructure builders between the application layer and the protocol layer. They are not base-layer chain founders, not top-tier research protocol inventors, and not NFT operators who built fame through celebrity-led narrative arbitrage. What they actually built is a service layer that turns difficult, brittle, slow, self-maintained IPFS and off-chain content infrastructure into something developers and enterprises are willing to pay for directly. Kyle’s role was to frame a technically neglected problem as a company big enough for capital and customers to care about. Matt’s role was to compress developer pain into something usable. Pinata is remembered not mainly because it spoke beautifully, but because it repeatedly landed on real infrastructure gaps. Compressed into a timeline, the story looks like this. In 2015–2016, Kyle encountered Bitcoin and Ethereum. In April 2017, he quit his job and entered crypto full time, building networks through hackathons. In 2017, he also built a blockchain meetup in Omaha and began collaborating with Matt. In 2018, the two explored chain applications through consulting and prototyping, then built the Pinata prototype at ETHBerlin and won the hackathon. From 2018–2020 they struggled through the crypto winter. In 2021 they raised a seed round. In 2022 they completed the Series A and accelerated alongside NFT infrastructure demand. In 2023 they sunset several private features. In 2024 they launched the Files API and a new file management layer. In 2025 they rebuilt the V3 API, launched Private IPFS and x402-based monetization, and also went through a significant outage. In 2026 they pushed Pinata Agents, OpenClaw/Hermes support, and agent templates much more aggressively into the foreground. Final conclusion. If you want a one-pass understanding of Pinata and its founders, the essence is this: Kyle Tut’s key strength was turning an unfashionable and difficult infrastructure problem into a story that capital, customers, and builders would all buy into, despite coming from a non-technical background. Matt Ober’s key strength was turning the kind of complexity that developers hate into products that are more usable and scalable. Pinata’s success does not come from embodying some pure ideological Web3 ideal; it comes from solving a very concrete problem—how off-chain data stays alive, stays fast, and remains callable by applications and machines. Its controversies come from the same realism: the more useful it becomes, the more it inherits the classic burdens of infrastructure companies—managed services, migrations, hidden dependencies, outages, and the centralization tension that comes with making decentralized systems actually usable.