TheFoundory

When a human employee joins a company, there is a process. Background check. Access controls. Permission levels. Regular audits. A clear record of what systems they can touch and what they cannot. When an AI agent joins a company, most of those guardrails don't exist. Enterprises are deploying AI agents across customer service, sales, finance, legal, and engineering workflows — often faster than their security teams can track. These agents have access to internal systems, customer data, and external APIs. They can take actions. Send communications. Process transactions. And in most organizations, nobody has a complete view of what they're doing, what vulnerabilities they carry, or whether they've been compromised. Straiker describes itself as "the agentic security company." Their platform discovers every AI agent operating inside an enterprise environment, tests those agents for vulnerabilities and misuse potential, and monitors them continuously in production. The company has grown 15x in revenue. On June 29, 2026, they raised a $64 million Series A led by Marathon Management Partners, with participation from Citi Ventures, Illuminate Financial, Workday Ventures, Bain Capital Ventures (BCV), and Lightspeed India. Total funding: $85 million. The investor list is itself a signal. Citi Ventures and Workday Ventures are strategic investors — financial services and enterprise HR — precisely the sectors deploying the most AI agents in the most regulated environments. Security used to mean protecting your perimeter from external attackers. In 2026, it also means governing the AI workforce you just hired. Straiker is building the infrastructure for that. #Straiker #AIAgentSecurity #CyberSecurity #StartupFunding #TheFoundory

Everyone is talking about GPUs. Nobody is talking about the real bottleneck in AI data centers. It's power. Here's a number that doesn't get enough attention. A single NVIDIA GB200 NVL72 rack — the kind being deployed in hyperscale AI data centers right now — consumes up to 120 kilowatts of power. That's roughly equivalent to powering 40 average American homes. From a single server rack. As AI data centers scale from hundreds of racks to thousands, the challenge of delivering that power efficiently, reliably, and without massive energy loss becomes one of the most critical engineering problems in the entire AI infrastructure stack. Reed Semiconductor Corp. is solving it. The Warwick, Rhode Island-based company builds multiphase power controllers and server power delivery modules — the hardware that sits between the electrical grid and the AI chips, converting and regulating power at the precise voltages and speeds that modern AI accelerators require. It is unglamorous work. It is also indispensable work. On June 29, 2026, Reed Semiconductor Corp. announced the completion of a $100 million funding round — oversubscribed, with participation from a number of leading global semiconductor companies. The round had previously been reported at a smaller size; the final close came in larger due to investor demand. The investor profile here is telling. This is not a consumer VC round. Global semiconductor companies wrote checks because they understand exactly what the power delivery bottleneck means for their own roadmaps. Every AI chip is only as useful as the power system that feeds it. Reed is building that system. #ReedSemiconductor #AIInfrastructure #PowerDelivery #Semiconductors #TheFoundory

Chamath Palihapitiya has had many titles. Facebook executive. Social Capital founder. SPAC sponsor. All-In podcast co-host. On June 29, 2026, he added a new one: full-time CEO of 8090 Solutions Labs. The company closed a $135 million Series A led by Salesforce Ventures, with participation from Jeffrey Katzenberg's WndrCo, David Sacks' Craft Ventures, and fellow All-In hosts David Friedberg and Jason Calacanis. Angel investors included Palo Alto Networks CEO Nikesh Arora and Quora CEO Adam D'Angelo. 8090 Solutions's product is called Software Factory. The name is deliberate. Most enterprise software today is built by teams of engineers and AI agents working in parallel — with no single view of what's being changed, by whom, why, and whether it's going to break something. The larger the organization, the worse the problem. Healthcare systems, aerospace contractors, financial institutions, and US government agencies are running mission-critical software that nobody has a complete map of. Software Factory brings people and AI agents into a single governed workspace — connecting business intent, architecture, code, testing, and production maintenance into one auditable pipeline. The result: software that can be changed at speed without pulling apart at the seams. EY deployed the platform internally and reported a 70% increase in development productivity and delivery speeds up to 80 times faster. Salesforce closed 22,000 Agentforce deals in its last fiscal quarter and froze software engineer hiring because AI tools were delivering enough. They then led 8090's round. "AI can write code," Palihapitiya said. "The hard part of enterprise software is keeping fifty agents and a hundred engineers changing the same complex system every week without it pulling apart." Since leaving Facebook, he had been waiting for a foundational moment to return to operations. He said he is convinced what they are building now is more important than the early social media era. For a man who has been right about enough big bets to earn the benefit of the doubt — that statement is worth paying attention to. #8090Labs #Chamath #EnterpriseAI #StartupFunding #TheFoundory

A chip company just paid $7 billion to get out of the data center and into the physical world. The battle for Physical AI just got serious. For the last three years, most of the AI hardware conversation has been about data centers. GPUs. Training clusters. Inference servers. The compute that powers large language models in the cloud. onsemi just made a $7 billion bet that the next AI hardware battle is somewhere else entirely. The company agreed to acquire Synaptics in an all-stock deal valued at approximately $7 billion. Synaptics brings connectivity chips, edge processors, and human-machine interface technology — the hardware that sits inside devices that need to sense, interpret, and respond to the physical world. Robots that navigate warehouses. Cars that understand their environment. Industrial systems that detect anomalies on a factory floor. Medical devices that monitor patients in real time. All of these applications need AI at the edge — on-device intelligence that doesn't rely on a cloud connection. That's the gap Synaptics fills. The acquisition gives Onsemi a credible position in the markets where Physical AI will actually land: automotive systems, robotics, industrial hardware, and connected devices. For context: the global edge AI market is projected to grow from $22 billion in 2025 to over $100 billion by 2030. The companies that own the hardware layer in that space will have an enormous structural advantage. Physical AI is moving from research slides to billion-dollar acquisition targets. That is how you know a category is real. #PhysicalAI #Semiconductors #Onsemi #Synaptics #TheFoundory

Samsung GSG (Global Strategy Group) plans to invest 1,000 trillion Korean won — approximately $648 billion — in South Korea over the next ten years. The investment covers chip factories in the country's southwest region, AI data centers, next-generation batteries, and advanced displays. The announcement is expected to be made at a meeting with South Korean President Lee Jae Myung. The timing is deliberate. AI workloads are driving unprecedented demand for memory chips — specifically the high-bandwidth memory that large language models require to run efficiently. Samsung Electronics is one of the world's largest producers of both DRAM and NAND flash memory. As AI infrastructure spending accelerates globally, Samsung's position in the supply chain becomes strategically critical. But this isn't just a business decision. It's a geopolitical one. South Korea, like many nations, is treating semiconductor manufacturing as national infrastructure. The investment is designed to decentralize tech capacity, create domestic jobs, and position Korea as an essential node in the global AI hardware supply chain — not a dependent one. For the global startup ecosystem, the implications are real. More chip supply means more compute availability. More compute availability means lower inference costs over time. Lower inference costs mean more founders can build AI products that were previously too expensive to run. The $648 billion Samsung bet is not just South Korea's story. Every AI startup in the world benefits when the hardware supply chain expands. #Samsung #AIInfrastructure #Semiconductors #DeepTech #TheFoundory

China's most talked-about AI startup just raised $7.4 billion and is preparing to double its workforce. This is what AI at national scale looks like. When DeepSeek released its R1 model in early 2025, it sent shockwaves through Silicon Valley. A Chinese AI lab had built a frontier-level reasoning model — at a fraction of the cost US labs were spending. The release triggered a debate about whether America's AI dominance was as secure as investors assumed. NVIDIA's stock dropped billions in market cap in a single session. Since then, DeepSeek has moved fast. The company has now raised more than $7.4 billion at a valuation above $50 billion. According to the Wall Street Journal, it is preparing to double its workforce — hiring across engineering, legal, finance, HR, and research roles. This is no longer a research lab story. This is an execution story. DeepSeek's backing comes from major Chinese institutional investors, and its growth trajectory is increasingly tied to Beijing's broader strategic agenda around AI leadership. The hiring push signals a company preparing to compete at industrial scale — not just model benchmarks. For the global AI ecosystem, the signal is important. The US vs. China AI race is no longer just about which country has the best models. It's about which country can scale faster, hire more talent, and deploy AI into real enterprise and government workflows at speed. DeepSeek is making a very clear statement about which side of that race China intends to be on. #DeepSeek #AIChina #StartupFunding #AIRace #TheFoundory

Airwallex was founded in Melbourne in 2015 by four co-founders who were frustrated by how expensive and slow international payments were for businesses. The specific pain point: running a small coffee shop that imported supplies from China. Every international transfer came with hidden fees, poor exchange rates, and settlement delays that made cash flow management a nightmare. The founders couldn't believe there wasn't a better system. So they built one. Today, Airwallex operates a full financial infrastructure platform for businesses global payments, multi-currency accounts, corporate cards, expense management, treasury operations, and now AI-powered finance tools. Customers range from startups to large enterprises operating across multiple countries. Revenue has been growing strongly. The company is active in over 150 countries and processes billions in payments annually. On June 26, 2026, Airwallex raised $320 million in fresh funding at an $11 billion valuation a 38% jump from its previous private valuation. The round was reported by Reuters. The capital will fund global expansion and deeper integration of AI across payment workflows, treasury management, and financial operations. Airwallex is not an AI company in the traditional sense. But AI is becoming the layer that makes its financial infrastructure smarter — predicting cash flow needs, flagging anomalies, automating compliance checks, and giving finance teams real-time visibility across currencies and accounts. Four co-founders. One frustrating coffee shop problem. An $11 billion company serving businesses in 150 countries. The best startups still start with the most personal problems. #Airwallex #Fintech #StartupFunding #CrossBorderPayments #TheFoundory

Solar-powered AI collars for cattle. Virtual fences managed from a smartphone. Just raised $220M — the largest VC round in New Zealand history. Farming has barely changed in a century in one fundamental way: you need physical fences to keep animals where they belong. Fences are expensive to build. Expensive to maintain. Inflexible. And they tell you nothing about the health of the animal inside them. Halter looked at that reality and built a completely different system. Each collar is solar-powered and fitted with GPS, accelerometers, and audio components. The collar monitors the animal's location, movement patterns, and health signals in real time. When an animal approaches a boundary, the collar emits a gentle audio cue to guide it back — a virtual fence, enforced by sound rather than steel wire. The farmer manages everything from a smartphone. Move a herd to a new paddock. Monitor individual animal health. Get alerts when something looks wrong. No physical fences required. The result: farmers report measurably better pasture management, improved animal health monitoring, and significant labor savings. Halter started in New Zealand — a country whose entire economy is partly built on livestock farming — found deep product-market fit, and has been expanding internationally. On June 5, 2026, Founders Fund led a $220M Series E at a $2B valuation. Blackbird, DCVC, BOND, and Bessemer also participated. The round nearly doubled Halter's valuation from just nine months prior. This is the largest venture capital raise in New Zealand's history. Not an AI lab. Not a defense tech company. A cattle collar startup from the southern hemisphere. Sometimes the biggest opportunities are the most overlooked ones. #Halter #AgriTech #StartupFunding #FoundersFund #TheFoundory

Five years ago, Polymarket was a niche crypto project where a small community of traders bet on election outcomes and random world events. Today, Reuters confirmed that Polymarket has surpassed $1 billion in annualized revenue — a milestone reached just six weeks after the company opened its US exchange to domestic customers. The platform lets people buy and sell contracts tied to real-world outcomes: who wins the FIFA World Cup, whether the Strait of Hormuz closes, when OpenAI goes public, what the Federal Reserve does next. Every market resolves based on what actually happens. The numbers tell the scale of what has happened to this category. Polymarket has recorded nearly $39 billion in trading volume in the US alone in 2026. New market launches have hit consecutive monthly highs for over a year. NYSE parent Intercontinental Exchange committed $2 billion to the platform — one of the most significant endorsements a traditional financial institution has ever given to a crypto-native startup. The reason prediction markets are growing this fast is not complicated. When real money is at stake, traders have every incentive to get the answer right. Aggregated predictions with financial stakes have consistently outperformed polls, expert panels, and institutional forecasts on everything from election outcomes to corporate announcements. Polymarket is not a gambling site. It is a real-time intelligence layer — one that financial institutions, hedge funds, and policy researchers now take seriously. A company that started as a niche corner of crypto is now generating $1 billion a year. And the US market just opened. #Polymarket #PredictionMarkets #Fintech #StartupFunding #TheFoundory

Sam Altman was handed two options: list now below $1 trillion, or wait until 2027. He called a lower valuation a "nonstarter." SoftBank lost $38 billion in a single day. OpenAI filed its confidential S-1 with the SEC on May 22, 2026. The filing was announced publicly on June 9. At the time, the company's private valuation stood at $852 billion — set in March when it closed a $122 billion funding round co-led by SoftBank, Amazon, and Nvidia. The plan was to go public in late 2026. Then the New York Times reported what was actually happening in the boardroom. Advisers presented Altman with a stark choice: accept a sub-$1 trillion valuation and list this year, or hold out for the trillion-dollar number and wait until 2027. Altman's answer, according to people familiar with the discussions: any cut to $1 trillion is a nonstarter. The market responded immediately. SoftBank shares fell as much as 14% in Tokyo trading before closing down more than 12% — wiping roughly $38 billion in market cap in a single session. SoftBank holds a massive position in OpenAI and has a $40 billion bridge loan due March 2027. A delayed IPO creates real pressure on that timeline. The financial case for waiting is real. OpenAI generated $13 billion in revenue in 2025 and is now running at $2 billion per month in 2026 — growing at roughly four times the pace Alphabet and Meta sustained at comparable stages. Another two quarters of growth significantly strengthens the IPO story. But the audited 2025 numbers also showed a $38.5 billion net loss. Reaching $1 trillion would place OpenAI alongside NVIDIA, Apple, Microsoft, and Alphabet — the only companies to have sustained that threshold in public markets. Prediction market Kalshi now gives 59% odds that OpenAI officially announces an IPO by March 2027. Earlier contracts had a 2026 listing at 30-40%. Altman is not rushing. He never said he was. #OpenAI #IPO #SamAltman #SoftBank #TheFoundory