TAIR — Tech & AI Report
Edition: May 15, 2026 • 7 Stories • Samwise Agency
Technology • AI • RoboticsJury to Decide Whether OpenAI Betrayed Its Founding Mission in Musk v. Altman
Elon Musk’s high-stakes legal battle against Sam Altman entered closing arguments this week, with the jury set to decide whether Altman and OpenAI breached a foundational agreement to develop artificial general intelligence for the public good. Musk claims OpenAI’s shift to a for-profit model betrayed its nonprofit mission — the very mission he helped fund and establish. Altman and OpenAI counter that the organization evolved appropriately to compete at the frontier of AI. The case hinges on contract interpretation and whether informal commitments made in OpenAI’s earliest days carry legal weight. A ruling could reshape governance norms across the AI industry.
Sources: TechCrunch
Cerebras Raises $5.5B in Blockbuster IPO, Opening at $185 to Kick Off 2026 IPO Season
AI chip specialist Cerebras Systems completed a blockbuster IPO this week, raising $5.5 billion and opening trading at $185 per share on Nasdaq — setting a jubilant tone for tech public offerings in 2026. The company, which designs wafer-scale chips optimized for AI inference workloads, saw share prices surge past $300 by day’s end, generating extraordinary returns for early backers including Benchmark Capital. Cerebras has positioned itself as an alternative to Nvidia’s dominant GPU ecosystem, arguing its WSE-4 chip delivers superior performance-per-dollar for large language model inference. The IPO marks a significant validation of purpose-built AI silicon as a distinct market segment.
Sources: TechCrunch
Cisco Cuts Nearly 4,000 Jobs to Fund AI Push While Reporting Record Quarterly Revenue
Cisco announced plans to eliminate nearly 4,000 jobs — approximately five percent of its global workforce — while simultaneously reporting record quarterly revenue, underscoring the industry-wide tension between AI-driven efficiency gains and traditional employment. The cuts are part of Cisco’s accelerated pivot toward AI infrastructure products, including networking gear optimized for hyperscale data centers and AI training clusters. CEO Chuck Robbins framed the layoffs as necessary to reallocate resources toward higher-growth segments. Despite the headcount reduction, Cisco raised its full-year guidance, citing strong demand for its Ethernet-based AI networking solutions as cloud providers race to expand AI compute capacity.
Sources: TechCrunch
Stealth Startup Recursive Superintelligence Raises $650M to Build AI That Designs Itself
A stealth startup called Recursive Superintelligence emerged from hiding this week, disclosing a $650 million seed round to build AI systems that design and improve themselves without human intervention. Founded by veterans of major AI labs, the company argues that the next leap in AI capability won’t come from scaling hardware or datasets, but from AI systems that autonomously architect their own successors. The announcement reignites longstanding debates about AI safety guardrails and whether recursive self-improvement is controllable. TechCrunch notes the company expects to have products ready within quarters, not years — a timeline that has alarmed alignment researchers and excited investors in equal measure.
Sources: TechCrunch
Wirestock Raises $23M to Supply Rights-Cleared Multimodal Training Data to AI Labs
Creative content platform Wirestock has raised $23 million to expand its human-generated multimodal dataset business, supplying curated image, video, and audio training data to AI labs struggling with data quality and copyright compliance. As frontier model developers face growing legal scrutiny over training data provenance, Wirestock offers a licensed, creator-compensated alternative to web-scraped datasets. The company claims its data significantly improves model fine-tuning outcomes across creative tasks. The funding round signals investor conviction that proprietary, rights-cleared training data is a durable moat in the AI stack — separate from model weights and compute — particularly as copyright litigation against AI companies continues to escalate.
Sources: TechCrunch
igus RBTX Marketplace Brings Low-Cost Humanoid Robots and AGVs to Factory Floors
German motion-plastics company igus showcased its RBTX platform at Hannover Messe 2026, positioning the marketplace as a one-stop procurement hub for low-cost humanoid robots and autonomous guided vehicles targeted at factory deployments. RBTX aggregates modular, plastic-component robot parts from multiple manufacturers, enabling smaller manufacturers to assemble capable automation solutions at a fraction of traditional robotics costs. The platform now includes several humanoid robot kits priced aggressively to compete with established industrial robot vendors. igus argues that commodity humanoid hardware, rather than expensive proprietary systems, will drive the next wave of factory automation — particularly in industries where labor costs are high and tasks are physically demanding.
Sources: Robotics & Automation News
Ai2 Releases MolmoAct 2: Open-Source Robotics Foundation Model Beats Proprietary Systems
The Allen Institute for AI (Ai2) released MolmoAct 2, an open-source robotics foundation model paired with the largest publicly available dataset for two-armed robotic manipulation — over 700 hours of coordinated task demonstrations. The model runs up to 37 times faster than its predecessor, outperforms leading proprietary robotics systems on industry benchmarks, and handles real-world tasks without per-task retraining. Designed to push robotics beyond controlled laboratory conditions, MolmoAct 2 is already being piloted at Stanford School of Medicine in CRISPR gene-editing workflows. Ai2’s open-source release directly challenges the proprietary data moats of commercial robotics AI vendors and accelerates broader research access to capable manipulation models.
Sources: Robotics & Automation News
⚡ Tech Pulse
Top Frontier Models (Arena Elo): GPT-5 (1,561) | Claude Opus 4.6 (1,503) | Gemini 3.1 Pro (1,494)
Top Open Source Models (SWE-Bench Verified): DeepSeek V4 Pro (80.6%) | MiniMax M2.5 (80.2%) | Kimi K2.6 (78.0%)
Top Small Models (15–50B, SWE-Bench Verified): Mistral Medium 3.5 (77.6%) | Kimi K2.5 (76.8%) | Qwen 3.5 (76.4%)
Top Edge Models (0–15B, tokens/sec): IBM Granite 3.3 8B (477 t/s) | Gemini 2.5 Flash-Lite (468 t/s) | Gemma 3n E4B (high-eff)
AI Leaders: Nvidia $5.2T | Alphabet $4.2T | Microsoft $3.2T
Robotics Leaders: ABB $190.9B | Intuitive Surgical $160B | Fanuc $38.9B