Investor Kotu is creating a joint venture with chip startup Maddox to finance and secure manufacturing capacity at fabs like TSMC. This innovative deal structure shows that for deep tech, venture capital is evolving beyond pure equity investment into actively de-risking critical supply chain bottlenecks for portfolio companies.
The perception that AI currently favors hackers over defenders is causing CISOs to increase cybersecurity budgets by 20-30%. This spending surge is a short-term reaction to catch up with new threats. Success is measured not by preventing breaches, but by a massive increase in newly discovered vulnerabilities found by defensive AI tools.
Foundational AI labs are now offering 'red and blue agents' capable of autonomously hacking systems to find flaws. This directly threatens the business model of established vulnerability scanner companies like Tenable and Rapid7, forcing them to quickly integrate LLMs or risk being displaced by more effective, model-native solutions.
Voice AI startup Eleven Labs, now profitable with $600M ARR, is targeting a 2028 IPO. This move signals the maturation of the AI application layer. Its public offering will be a key test of whether specialized AI companies can build a defensible business on top of foundational models rather than being subsumed by them.
OpenAI now publicly labels its Astra model as AGI, a significant communications shift. This is enabled by a revised Microsoft agreement and a broader industry re-focus. The new existential threat is 'super intelligence,' making the AGI label a less controversial and more viable marketing term.
Anthropic is hiring to build its own payments infrastructure for billing, fraud, and treasury. This move highlights how the complexity of usage- and token-based AI pricing models makes off-the-shelf solutions from vendors like Stripe insufficient, forcing AI leaders to become FinTech companies themselves.
After a rocky start, AWS's AI platform Bedrock is winning customers back from Microsoft Azure. The turnaround wasn't driven by a new model, but by a small engineering team that redesigned Bedrock's backend ('Project Mantle') to handle the uniquely unpredictable and resource-intensive 'spiky' nature of AI inference tasks.