Lumentum Holdings Inc. says its optoelectronic parts are “completely sold out” through almost 2029, driven by demand from tech companies building faster AI data centers. The claim comes from the company, and the report gives no customer names, order values or capacity figures.
The Information reports that the widely circulated figure of $70 billion in OpenAI annualized revenue was far off, and that the company is generating closer to $50 billion. After a Financial Times report of the lower number on Thursday, shares of chip firms and neoclouds including Nvidia, CoreWeave and Nebius fell. The piece argues that multiplying one month's revenue by 12 is a poor substitute for a fuller picture of a company's performance.
Bloomberg Intelligence analyst Mandeep Singh noted that questions remain over how OpenAI accounts for its revenue, while booming token growth points to solid demand for AI infrastructure. The analysis suggests token usage can serve as a demand gauge even if revenue figures are restated.
In Canada, Okanagan Specialty Fruits is testing a tractor-mounted camera system from local firm Vivid Machines that uses AI to spot buds, flowers and fruit as it moves through orchards, estimating yields and optimal harvest windows; experts say accuracy depends heavily on the quality of historical data and requires per-farm customisation rather than a generic model. In the UK, FruitCast gathers drone, smartphone or machinery imagery to help growers track ripening of strawberries, raspberries, blackberries, blueberries and tomatoes, with co-founder and COO Raymond Martin noting that strawberries must be picked as soon as they ripen. The lighter-weight Fruit Forecast app has growers film 50 representative plants each week and combines that with weather forecasts, variety data and fruit weight to generate four-week harvest projections; it is already on Android and Apple app stores and offers an API.
MIT Technology Review will host a subscriber-only online roundtable on October 16, 2026, with AI reporter James O'Donnell interviewing Samuel King, a Stanford bioengineering PhD candidate and Innovators Under 35 honoree. The listing recalls that in 2025 King used a generative AI model to propose genetic blueprints for microscopic viruses, noting this is not yet AI-generated life but "that could be next." The conversation will cover his work and "new ways of seeing biology."
Anthropic announced Anthropic Cyber Mission, described as a long-term commitment to use its tools, research and resources to help defenders protect critical infrastructure and open-source software. Under it, the Critical Infrastructure Defense Program (CIDP) will give defenders guarding the operational technology behind power grids, water systems, transport networks and government systems access to frontier models, embedded engineers and threat research. The report gave no timeline, participating organizations or funding figures.
An article republished by 36Kr from 动脉网 tallies more than $3B in direct and indirect funding over the past year or so into "physical AI" for AI4S — systems that let AI operate instruments and robots in real wet labs. It names Medra, 津渡生科, 术也科技, 合碳智能, 宇道生物, 予路乾行, 深度细胞, Phyraxis AI and Transfyr, alongside NVIDIA's joint lab with Eli Lilly, a Simcere–Schrödinger drug discovery deal and Anthropic's open MHS model-hardware standard with HHMI Janelia. The piece argues AI drug discovery's bottleneck is no longer algorithms or compute but wet-lab data and hardware, with failed-experiment data locked inside companies creating a structural mismatch.
Eliyan, a five-year-old startup whose technology lets separate chips inside an AI system communicate faster and more efficiently, has told some investors in recent days that it received a takeover bid, according to The Information. The company, valued at $1 billion in July, has worked with an investment bank since the summer and is likely aiming for a $3 billion valuation. SoftBank-owned Arm, an existing investor, held early talks about a possible purchase but did not make a bid and would be unlikely to do so at that valuation, a person familiar said.
TechRadar reports that John M. Wilson, Google's former global head of design and construction, has joined Anthropic as a technical staff member overseeing EMEA capacity delivery. He spent six years at Google covering data center development and subsea infrastructure programs, and previously led European design and construction at Amazon, after roles as chief of staff in the British Army and infrastructure director at NATO. The report also lists several other former Google infrastructure specialists now at Anthropic across data center design, electrical systems, security, operations, energy, capacity delivery and compute finance.
A-share optical communication stocks pulled back on October 8. Dongshan Precision said in an institutional research session that the widely cited claim of a "65% US-origin materials ratio" is an unofficial market rumor, and that it has already arranged multi-region capacity and diversified supply-chain options. As AI model clusters keep scaling, traditional bandwidth and power bottlenecks are becoming more acute, pushing optical interconnect into parallel iteration across pluggable, NPO and CPO routes and accelerating commercialization, with optical firms moving further upstream.
According to Securities Times data cited by 36Kr, Chinese AI agent concept stocks fell an average 20.63% so far this year as of the Oct. 8 close, badly trailing the Shanghai Composite, though seven names gained more than 10%. Thirty-nine of the stocks drew "positive" ratings from at least five institutions, with Hisense Visual Technology mentioned in the most broker reports (28); the company says it has built an integrated "device-cloud-chip" AI stack and shipped terminal applications such as the Xiaoju Super Agent and multimodal Xiaoju screen Q&A. The report frames the AI narrative as shifting from a model-capability race to Agent commercialization.
That's everything.