Google kicked off its I/O 2026 developer conference with one of the year's biggest AI moves: the launch of Gemini 3.5 Flash — the first model in the Gemini 3.5 family — alongside Gemini Omni, a unified multimodal platform for generating video, images, and audio from any input type.
The new model, available immediately to all Gemini app users, delivers roughly 4× faster output than comparable frontier models at less than half the cost. It supports a 1-million-token context window — enough for entire books — and up to 65,000 output tokens in a single response.
Why it matters
Gemini 3.5 Flash isn't a routine model upgrade. Google is positioning it as an "agentic" model — one that can autonomously execute complex, multi-step workflows like building a full application, analyzing documents, or operating external tools, rather than just answering questions.
On benchmarks, the new model surpasses Gemini 3.1 Pro on coding and agentic AI metrics, scoring 76.2% on Terminal-Bench 2.1 and 83.6% on MCP Atlas. Alongside it, Google introduced Gemini Omni — a "world model" that generates physically accurate video from any input, with conversational editing capabilities. It first rolls out to AI Plus/Pro/Ultra subscribers.
The announcements were rounded out by Gemini Spark, a proactive 24/7 personal AI agent. The full Gemini 3.5 Pro is slated for June.
Lenovo sets records — AI revenue soars 84%
Beyond Google's announcements, the trading week opened with strong quarterly results from Lenovo (LNVGY). The Chinese PC maker reported record Q4 revenue of $21.6 billion, up 27% year-over-year — its fastest growth in five years. The stock jumped roughly 15% on the release.
The headline number: AI-related revenue surged 84% YoY, now accounting for 38% of total group revenue. Lenovo maintained a global PC market share of 24.4%, with premium PC shipments up 29% — a category that includes AI PCs.
Nvidia goes quantum
Nvidia's venture arm NVentures invested in French quantum startup Alice & Bob, which previously closed a €100 million Series B round. Alice & Bob is developing a proprietary "cat qubit" architecture designed to suppress bit-flip errors and reduce error-correction overhead — one of the biggest challenges in practical quantum computing. The investment expands a deep technical collaboration that includes NVQLink, an architecture integrating quantum processing units with Nvidia GPUs.
European AI stocks shine despite macro gloom
Despite economic headwinds from the Iran conflict and rising energy prices, European AI stocks are showing remarkable resilience. A basket including ASML, Infineon, and STMicroelectronics has rallied roughly 20% since early April. ASML recently raised its full-year 2026 sales forecast to €36–40 billion, citing surging AI spending.
AMD commits $10 billion to Taiwan ecosystem
AMD announced more than $10 billion in investments across Taiwan's semiconductor ecosystem, aimed at expanding advanced packaging capacity and AI chip production. The new Instinct MI450X GPU and Helios rack-scale platform are on track for volume production in the second half of 2026.
The bottom line
This week serves as a reminder that the AI industry is in the midst of a historic investment wave — from the largest models (Google, OpenAI) to infrastructure (ASML, AMD, TSMC) and consumer hardware (Lenovo). Google, with the dual punch of Gemini 3.5 Flash for everyday workloads and Omni for multimodal creation, has set a new competitive bar. The big question: how quickly will these new models capture real usage share — and how will OpenAI respond?
