A French AI-Networking Deal Lands Amid the Sell-Off
While US chip stocks have been pummelled — the Philadelphia Semiconductor Index (SOXX) down 25% from its highs and leveraged semiconductor ETFs down 62% from their peaks, according to Via News's tracked market data1 — European chip suppliers kept signing deals through the same window. On July 28, 2026, Grenoble-based Kalray (Euronext Growth Paris: ALKAL), a specialist in hardware and software for intensive computing and data infrastructure, announced a collaboration with Bull to develop next-generation high-speed networking technology for AI and high-performance computing infrastructure, including compatibility with the Ultra Ethernet standard.2 "Cet accord avec Bull confirme une nouvelle fois la pertinence de notre vision technologique et la qualité des innovations développées par nos équipes," said Kalray CEO Éric Baissus.2 The companies say the partnership aims to develop solutions enabling the next generation of AI and HPC and to create long-term value for customers worldwide.2
It is a small, publicly-listed European name making a deal, not a headline-grabbing mega-merger — but it is precisely the kind of activity that illustrates the gap this correction has opened: infrastructure buildout continuing at the working level even as public equities in the sector have been marked down sharply.
The Global Deal Flow Kalray's Announcement Sits Inside
The networking tie-up follows a week of much larger AI-infrastructure agreements elsewhere in the chain. On July 24, 2026, Nvidia's Jensen Huang joined South Korean President Lee Jae Myung, Sam Altman, and the heads of Samsung, SK Group, Hyundai Motor and Naver in San Francisco, where roughly $950 billion in new AI agreements were signed.3 Among them, Nvidia and SK Hynix expanded their partnership; "the expansion will include a co-develop opportunity for us on the next-generation SK Hynix AI memory, and this will help us secure a stable supply of HBM memory," said Raj Mirpuri of the arrangement.3 Days earlier, on July 23, 2026, AMD and Cerebras announced a joint ultra-low-latency AI inference solution combining AMD Instinct GPUs with the Cerebras Wafer-Scale Engine, with Cerebras planning to deploy AMD's Helios systems in its data centers and the joint offering reaching Cerebras Cloud in the second half of 2026.4
On the manufacturing side, GlobalFoundries signed a letter of intent with the US Department of Commerce on July 29, 2026, under which the CHIPS Research and Development Office is expected to award the company $300 million to advance next-generation silicon photonics — the optical technology used to move data at high speed in AI and high-performance computing data centers.5 AMD's Mark Papermaster welcomed the move: "As AI systems scale, moving data efficiently is as critical as increasing compute performance. Silicon photonics and advanced packaging will be key to delivering the bandwidth, energy efficiency, and system-level connectivity required for the next generation of AI cluster infrastructure."5 It is the same connectivity problem Kalray and Bull are working on from the European side, in the same week.
What the Underlying Companies Are Actually Reporting
Away from the deal announcements, the fundamentals in AI-adjacent chip suppliers still look strong. Vicor Corporation, whose power components feed AI computing infrastructure, reported trailing twelve-month revenue of $471.7 million and net income of $136.7 million, with a market capitalization of $11.3 billion as of the July 6, 2026 market close.6 Taiwan's Phison, in a June 2, 2026 collaboration with Intel announced at Computex, is extending local memory for AI PCs; its Pascari aiDAPTIV technology is meant to let larger mixture-of-experts models and agentic applications run locally. "AI PCs are evolving into platforms for more sophisticated local AI workloads, including agentic applications and larger MoE models that place increasing demands on memory capacity and responsiveness," said KS Pua.7
Smaller suppliers are showing the same demand pull. Allegro MicroSystems, which describes itself as leveraging more than three decades of expertise in magnetic sensing and power ICs to propel electrification, automation, AI data center, and robotics applications forward, appointed semiconductor-industry veteran Brian White to its board on June 18, 2026.8 "Brian's public-company CFO perspective, semiconductor industry experience and governance background make him a strong addition to our Board," said Allegro's Joseph Martin.8 At the smaller end, Florida-based Solitron Devices reported fiscal 2027 first-quarter net sales up 101% year-over-year to roughly $5.44 million, with backlog up 28% to $23.34 million; the company said net income was impacted by a $0.33 million increase to contingent consideration tied to a large order received after quarter-end, and that it expects "sales to continue to be at this level or greater for the remainder of the 2027 fiscal year."9
Where Sentiment and Fundamentals Diverge
Not every recent data point points the same way. Amkor Technology guided for third-quarter net sales of $1.95 billion to $2.05 billion — weaker than the $2.11 billion analyst consensus — contributing to a broader rout in chipmaker and AI-infrastructure stocks that dragged the Nasdaq 100 to a 2.75-month low on July 28, 2026, even as the S&P 500 and Dow Jones Industrial Average were roughly flat to higher that same day.10 Consolidation is also reshaping the sector's structure: Skyworks Solutions and Qorvo announced their expected combined leadership team on July 28, 2026, ahead of completing their previously announced merger. "Today's announcement reflects the strong partnership that has shaped our integration planning efforts from the very beginning," said Skyworks's Bob Bruggeworth.11
Volatility in the sector's public equities has been extreme even by chip-cycle standards. ProShares' 2x-leveraged Ultra Semiconductors fund (USD) was up about 69% year-to-date as of June 5, 2026 — from $52.45 at year-end — and 196% over the trailing year, from $29.94 a year earlier.12 That degree of leveraged upside is the mirror image of the subsequent drawdown Via News's tracked data shows across leveraged semiconductor products, down 62% from their highs.1 Micron Technology's stock offers one possible read on where the correction stands: after falling nearly 40% from its all-time high beginning in July 2026, Micron bounced 10% off its lows by August 9, 2026, though it remained roughly 30% below its peak.13 Via News's analysis of this pattern — deal flow and guidance continuing to firm even as leveraged trading vehicles round-trip through outsized gains and losses — is consistent with a late-cycle correction rather than a reversal in underlying AI infrastructure demand, though this reading is our own interpretation of the tracked data, not a confirmed outcome.
What to Watch
For European readers, Kalray's Bull partnership is a live test of whether smaller listed European suppliers can convert AI-networking demand into revenue while their larger US and Asian counterparts absorb the bulk of both the deal flow and the equity volatility. On the US supply side, Vicor's lead computing customer is scheduled to transition from Gen 4 to Gen 5/VPD technology in the second half of 2026, with a ramp expected before year-end — a concrete near-term signal for power-delivery demand tied to next-generation AI compute.6 Whether Amkor's softer guidance is an isolated packaging-demand wrinkle or an early sign of broader order softening, and whether Micron's bounce off its lows holds, are the two clearest near-term tests of whether this is a correction inside a supercycle or the start of something else.
