Three developments · Growth, technological sovereignty and AI oversight · About 2 minutes
Euronews · Published October 13, 2026, 04:49 CEST
The IMF’s new outlook raises expected global growth this year to 3.1%, still below last year’s 3.5%. Its explanation is more revealing than the headline: AI investment is helping offset the energy shock from the Middle East war, but the benefits and costs land in different places. Economies connected to technology supply chains gain; energy-importing and low-income countries face higher bills without an equivalent investment windfall. Eurozone growth is forecast at just 1.1%.
This is not a uniform recovery. It is a redistribution of momentum, with global inflation forecast at 4.7%. Nor does investment spending prove that AI has already delivered lasting productivity gains. The IMF warns that reassessing expected AI profits could abruptly weaken investment and tighten financial conditions. An economy supported by one concentrated boom remains vulnerable even when its aggregate numbers improve.
Reuters via MarketScreener · Published October 13, 2026, 04:31 CEST (October 12, 22:31 EDT)
Newly reported IDC figures put Nvidia at 34% of China’s AI-accelerator market in the first half of 2026, with Huawei at 28%. Nvidia’s unit shipments fell 44% year on year. The underlying measurements cover January–June; today’s news is their publication, not an overnight change in market share.
US export restrictions and Beijing’s support for domestic substitutes are reshaping an entire computing ecosystem. Non-GPU accelerators now outsell GPUs by unit volume, and IDC says competition increasingly concerns how chips work together in large clusters. The caveat matters: shipment share does not establish equal computing power or technological parity. But the data show that restricting access can also accelerate substitution and erode American suppliers’ commercial influence.
Reuters · Report dated October 13, 2026
Reuters today revealed negotiations in which Anthropic and OpenAI sought significant control over what outside evaluators could publish, according to two people familiar with the talks. Both publicly endorse independent assessment. Anthropic disputes the broad characterization and says established reviewers control publication; OpenAI says it seeks confidentiality redactions, not approval of entire reports.
The structural issue is who decides what the public gets to learn. Trade secrets deserve protection, but information relevant to safety can overlap with proprietary material. Without clear rules for access, redactions and disclosure, commissioning an outside review is not the same as accepting independent scrutiny. The negotiations—not last month’s safety pledges—are today’s new development.