Optical Stocks Sold Off on AI Funding Fears, Not New Company Data

Doximity and OpenEvidence have ended a contentious legal dispute over clinical artificial intelligence, with a federal judge dismissing the case with prejudice. That means the claims cannot be filed again in the same court. Each company agreed to pay its own legal costs, while other settlement terms were not disclosed.

OpenEvidence originally accused Doximity employees of posing as physicians and using provider identification numbers to gain access to its platform. The complaint alleged that users submitted prompts designed to reveal system instructions and proprietary information. Doximity denied wrongdoing and countersued for defamation and harassment, arguing that OpenEvidence had made misleading public comparisons and pressured employees.

The allegations changed during the litigation. OpenEvidence later dropped its trade-secret claims and focused on alleged violations of federal anti-hacking law and misuse of the platform. The settlement arrived as OpenEvidence faced a deadline related to internal Slack messages that Doximity said had been deleted. Neither company provided a detailed public explanation of the agreement.

The dismissal removes a legal distraction, but it does not establish that either side’s allegations were proven.

Both companies remain direct competitors. Doximity offers an AI-backed clinical assistant alongside its professional network for physicians, while OpenEvidence provides an AI medical search and reasoning platform for verified doctors. Doximity says its clinical AI products are used by more than 165 health systems. It also signed a partnership with Wolters Kluwer to bring UpToDate information into Doximity Ask.

For Doximity shareholders, the settlement reduces uncertainty around legal expense and management attention, but the financial effect cannot be measured because payment and conduct terms were not disclosed. The more important long-term question is whether Doximity can convert physician usage into durable revenue while maintaining accuracy, security and trust.

Doximity reported fiscal 2026 revenue of $644.9 million, up 13%, and free cash flow of $317.5 million, up 19%. Those figures provide a profitable base for investment in clinical AI. The settlement clears one obstacle, but competitive performance will be determined by adoption, product quality and customer retention rather than the court outcome alone.

Alaska Airlines has appointed Amaresh Siva as chief information and technology officer, bringing in an executive with experience across health care, retail and industrial companies. The role covers enterprise technology, data, digital products, cybersecurity and artificial intelligence, according to the airline.

Siva joins Alaska after serving in senior technology roles at CVS Health. Alaska said his career spans nearly three decades and also includes Target and GE. The breadth is relevant because airline technology touches reservation systems, operations, maintenance, payments, customer service and employee tools. Failures can interrupt travel, while well-designed systems can reduce costs and improve reliability.

The timing adds significance. Alaska Air Group is integrating Hawaiian Airlines, a process that requires systems, data and operating procedures to work together without disrupting customers. Technology leadership will influence how quickly the group combines platforms, protects sensitive information and delivers a consistent digital experience across brands.

Siva’s appointment is strategically relevant because Alaska’s integration challenge is as much about systems and data as it is about aircraft and routes.

The position does not create an immediate financial forecast. No compensation terms, cost-savings target or technology budget was disclosed in the reviewed materials. A senior hire can improve execution, but the effect must be measured through operational results rather than a résumé alone.

Investors should look for evidence in several areas: fewer technology-related disruptions, better on-time recovery tools, progress toward common customer platforms, stronger cyber controls and clear savings from replacing duplicate systems. AI initiatives should also be judged by observable outcomes such as faster service, better scheduling or lower support costs, not by the number of projects announced.

Alaska included the appointment in its third-quarter corporate update, confirming that Siva joined the executive committee. That gives the technology function direct representation in strategic decisions, which is important during a multi-year integration.

The hire is a sensible step for an airline with a complex transformation ahead. It is not, by itself, a reason to change earnings estimates. The investment value will emerge if Alaska converts Siva’s cross-industry experience into reliable systems, measurable integration progress and a better customer journey while keeping cybersecurity and operational resilience intact.

Hewlett Packard Enterprise’s third-quarter results showed how quickly the company’s scale has changed after the Juniper Networks acquisition. Revenue reached $12.2 billion, up 34% from a year earlier, while reported gross margin was 40.1% and operating margin was 11.4%.

The growth is meaningful, but it needs interpretation. Acquisitions can lift reported revenue before they improve earnings per share or free cash flow. Investors must separate organic demand from the contribution of Juniper, then compare the added profit with integration costs, financing and any new share count.

Management also said supply remained a constraint. That can be a positive demand signal when orders exceed available components, but it can also delay revenue, raise procurement costs and frustrate customers. The financial effect depends on whether HPE can protect pricing and deliver the delayed equipment without losing business.

HPE’s 34% revenue growth is impressive, but durable value will depend on margins, cash conversion and successful integration rather than sales alone.

The original downside analysis emphasized that HPE shares can be volatile and that historical selloffs have sometimes been severe. Historical drawdowns do not predict a new decline, yet they remind investors that hardware businesses are cyclical and sensitive to capital spending. The AI server opportunity adds growth, but it also exposes HPE to expensive components, fast product cycles and strong competition.

Juniper can improve the mix by adding networking software, recurring revenue and higher-value infrastructure. The strategic case is strongest if HPE can sell compute, networking and services together while retaining customers and reducing overlapping costs. Integration missteps would weaken that benefit and make leverage more important.

The next reports should answer three questions. How much of revenue growth is organic? Can gross and operating margins hold as constrained products ship? Is free cash flow improving after working-capital and acquisition-related demands? Clear progress on all three would make the recent growth more investable.

HPE is no longer only a mature enterprise-hardware story. It now has a larger networking platform and more exposure to AI infrastructure. That can raise the company’s growth ceiling, but it also raises the execution standard. Investors should require evidence that scale is producing better economics per share, not merely a larger income statement.

Shares tied to memory and storage fell as investors reconsidered how long the sector’s favorable pricing could last. The selloff followed concern around SK Hynix’s investment plans and the possibility that additional capacity could eventually weaken the shortage conditions that have supported prices.

That reaction captures the central tension in the memory cycle. Artificial intelligence servers require large amounts of high-bandwidth memory and storage, creating strong demand for producers across the supply chain. Yet the same high prices and margins that attract investors also encourage manufacturers to spend more. Because new fabrication capacity takes time to build and qualify, the market can move from shortage to surplus with a delay.

SanDisk enters that debate with powerful recent results. The company reported fiscal 2026 revenue of $20.25 billion, up 175%, while data-center revenue rose 437%. Management said higher pricing contributed to the increase. Those figures confirm that the current cycle is strong, but they also make future comparisons more demanding.

Memory investors are not only betting on AI demand; they are betting that supply will remain disciplined enough for prices and margins to stay elevated.

The read-through from SK Hynix is not mechanically negative. Spending on advanced memory may be necessary simply to serve contracted demand, and some capacity cannot easily shift between products. High-bandwidth memory also carries different technical requirements from standard NAND storage. Still, investors often sell the group when they see signs that industry capital spending could outrun demand growth.

For SanDisk, the most useful indicators are bit shipments, average selling prices, inventory and gross margin. Revenue alone cannot show whether growth comes from sustainable volume or a temporary pricing spike. For Micron and SK Hynix, the timing of new capacity and customer commitments will help reveal whether investment is defensive, growth-oriented or excessive.

The decline therefore does not invalidate the AI storage thesis. It shows that the thesis has moved into a more mature phase. Demand growth is already widely recognized, so the next source of upside must come from earnings that remain strong even as competitors add supply. If pricing softens faster than volumes expand, memory stocks can fall despite continued growth in AI infrastructure.

Nvidia’s announcement that SpaceXAI will adopt its Vera CPU gave the semiconductor sector another reason to price in expanding artificial intelligence infrastructure. Nvidia said the deployment would support agentic AI at massive scale and expand Grok infrastructure on the Vera Rubin platform. The announcement did not disclose a purchase volume or contract value.

The direct commercial link is therefore clearest for Nvidia. SpaceXAI selected Nvidia’s technology for a defined infrastructure plan, which can support future processor and platform demand if deployment proceeds. For SanDisk and Intel, the market reaction described by 24/7 Wall St. was a broader read-through rather than evidence of new contracts.

SanDisk can benefit when AI systems require more high-performance storage, and Intel can benefit if data-center spending broadens across processors, networking and systems. However, a customer’s decision to use Nvidia’s CPU does not specify whose storage or complementary silicon will be installed. Investors should avoid converting sector enthusiasm into an unsupported revenue forecast for every chip company.

The SpaceXAI announcement is a direct product endorsement for Nvidia, but only an indirect demand signal for SanDisk and Intel.

SanDisk’s own fundamentals provide a more concrete backdrop. The company reported fiscal 2026 revenue of $20.25 billion, up 175%, with data-center revenue up 437%, helped by pricing and demand. Those figures show that AI infrastructure spending is already affecting its business. They are stronger evidence than a single peer announcement, though such growth rates can also attract new supply and make future comparisons difficult.

The broader lesson is that AI hardware rallies often combine two different ideas: confirmed company-specific demand and expectations that spending will spread through the supply chain. Both can move prices, but they deserve different confidence levels. Nvidia has the confirmed platform selection. SanDisk has separate evidence of strong data-center growth. Intel’s benefit from this particular announcement remains less specific.

Investors should watch for purchase commitments, shipment schedules and customer disclosures that narrow the gap between enthusiasm and revenue. Without those details, the announcement supports the sector’s demand narrative but does not establish equal economic value for every stock that rose alongside Nvidia.