TECHDESK DAILY — Pre-Market Edition · Friday, November 13, 2026

Pre-market: prices are the Thursday November 12 close plus any pre-market quotes. · 115 stories reviewed

Top Stories

SpaceX developing custom Nvidia Vera Rubin rack dubbed Kessel for its orbital data centers ▲ LOW DISSEMINATION

SpaceX is prototyping a bespoke Nvidia Vera Rubin rack, nicknamed “Kessel,” for use in its planned orbital data centers. The design keeps the underlying configuration but swaps the PCB to suit space‑grade requirements【Data Center Dynamics】(https://www.datacenterdynamics.com/en/news/spacex-developing-custom-nvidia-vera-rubin-rack-dubbed-kessel-for-its-orbital-data-centers-report/). This effort aligns with Teledyne e2v’s expansion of space‑qualified DDR4 memory, underscoring a broader push for advanced AI‑ready hardware in orbit.

Desk take: Space‑grade AI compute is becoming a competitive moat; watch for supply‑chain implications for Nvidia and memory vendors.

Teledyne e2v Expands Space DDR4 Memory Line With 8 GB and 16 GB NASA Grade Models ▲ LOW DISSEMINATION

Teledyne e2v announced new 8 GB and 16 GB NASA‑grade DDR4 modules aimed at the growing onboard storage needs of Earth‑observation, communications payloads, and AI workloads in space【SatNews】(https://satnews.com/2026/11/12/teledyne-e2v-expands-space-ddr4-memory-line-with-8-gb-and-16-gb-nasa-grade-models/). The launch complements SpaceX’s custom Nvidia rack, together bolstering the hardware stack for next‑gen orbital data centers.

Desk take: Expanded space memory capacity could lift demand for high‑performance chips; consider exposure to both memory and GPU suppliers.

Amazon Leo Scheduled to Launch 40 Satellites on ULA Vulcan Rocket UNDER-COVERED

Amazon Leo will deploy 40 broadband satellites aboard a United Launch Alliance Vulcan rocket from Cape Canaveral on Nov 13【SatNews】(https://satnews.com/2026/11/12/amazon-leo-scheduled-to-launch-40-satellites-on-ula-vulcan-rocket/) and 【NASASpaceflight】(https://www.nasaspaceflight.com/2026/11/amazon-leo-lv-01/). This launch adds to the emerging capability set for commercial LEO constellations, mirroring AST SpaceMobile’s push to build its own reusable launch vehicle.

Desk take: New launch cadence expands market for satellite‑based broadband; monitor downstream demand for ground‑segment hardware.

AST SpaceMobile presses case for developing own launch vehicle UNDER-COVERED

AST SpaceMobile executives claim they can field a reusable launch vehicle within two years, despite industry skepticism【SpaceNews】(https://spacenews.com/ast-spacemobile-presses-case-for-developing-own-launch-vehicle/) and 【Ars Technica】(https://arstechnica.com/space/2026/11/building-a-rocket-is-hard-ast-spacemobile-thinks-its-simple/). Their ambition parallels Amazon Leo’s upcoming Vulcan launch, signalling a broader trend of satellite operators seeking launch autonomy.

Desk take: Vertical integration in launch capability could reshape cost structures for satellite services; keep an eye on capital allocation.

Why Singapore keeps turning up in the fight over Nvidia chips and China ▲ LOW DISSEMINATION

Singapore barred Alice Huang from re‑entering the country amid investigations linking her to Nvidia‑related activities, with limited U.S. assistance reported【CNBC Technology】(https://www.cnbc.com/2026/11/13/nvidia-jensen-huang-alice-spy-megaspeed-.html). Her stakes in firms providing compute to Huawei highlight the geopolitical tug‑of‑war over Nvidia technology, echoing U.S. congressional moves to curb China’s access to certain Nvidia chips【Reuters】(https://news.google.com/rss/articles/CBMixAFBVV95cUxQOFBGZG9jUFFjX0tFY2tXSEVILWxGWU1rMzBvT1pTcUpqa2tEdEV0QWRybENzel9tU3hIQVhmTngyY3VGRGZZeGtCWDhEWm1oRXBzVWlGRnhwRzJrd0Z5bGZhUi1STHFLdFdLVnhoRUxMOW42MkNSSjNZX3JQR0o3OVpBSE10bE1kekw3OVVXYXVub0NzekN5NE4zb1JpRzB6Zjl2V1pTWFEteUQ1Q0hxd2lhLV9JLXloLVJqbWhUQ1pmN21H?oc=5).

Desk take: Geopolitical risk around Nvidia chips is intensifying; exposure to supply‑chain constraints in AI hardware warrants scrutiny.

SK Hynix, Intel explore shared memory to ease AI bottlenecks ▲ LOW DISSEMINATION

SK Hynix and Intel are collaborating on memory‑centric computing architectures that let AI servers share memory resources, addressing rising AI‑driven demand for capacity【DigiTimes】(https://www.digitimes.com/news/a20261113PD229/sk-hynix-intel-data-capacity-cxl.html). This effort dovetails with Intel’s broader AI ecosystem push, including a GPU alliance and low‑power DRAM strategies【DigiTimes】(https://www.digitimes.com/news/a20261113VL226/intel-gpu-alliance-president-design.html). The partnership reflects the same AI compute bottleneck theme highlighted in the edition’s AI‑centric hardware note.

Desk take: Shared‑memory solutions could mitigate AI server shortages; monitor joint roadmaps for potential upside in both memory and CPU stocks.

Google reworks TPU silicon into specialized systems for agents and security ▲ LOW DISSEMINATION

Google has repackaged its TPU 8i silicon into two new chips—TPU 8i Max and TPU 8tx—targeting long‑context AI agents and confidential inference workloads, respectively【Tom's Hardware】(https://www.tomshardware.com/tech-industry/artificial-intelligence/google-adds-two-new-specialized-chips-for-ai-agents-and-security-to-8th-gen-tpu-lineup-tpu-8i-max-puts-two-chips-on-one-package-for-long-context-work-tpu-8tx-swaps-arm-hosts-for-amd-epyc). This specialization aligns with the broader AI hardware acceleration trend across the sector.

Desk take: Google’s chip diversification may boost its AI service margins and create competitive pressure on Nvidia; assess impact on cloud AI pricing.

Under the Radar

• ▲ LOW DISSEMINATION — OpenAI Ex-CTO Mira Murati’s Thinking Machines Lab targets $40B valuation in Accel‑led round amid Nvidia alliance: The lab is seeking a $40 billion valuation as it raises a new round led by Accel, with Nvidia positioned as a strategic partner【https://news.google.com/rss/articles/CBMi8AFBVV95cUxQbVFTU3l4Q05lNVZ2X21PN3NNNjlZNzJsVmpZeGZlaWV1Uk5WcHM5dUhjTno4NjFyb3VzYl9VTkV4VTIxbFRvVnZ4MmZSUmM1SHExdFEwZ3ZsaHdXcmRIc19KZS1UUXh4dEcyaWpXcFg3MllYbmQyZDI2X0FqSExXZ2dMcDZvYVpJUWFLYUZoUGlrQnNid1NjRXJmSmFtV2d3eTM5NzYwUEd4ZEt2cnZEMjdTWkFIajJxVnlVVHlqdGlPYWtva1E2UWstdld6Tk42TkJJYmpxNWJRX1JQNEF0VkE1WHRLUnZiNFgtNWhNZGk?oc=5】. Desk take: Potential new AI challenger could reshape competitive dynamics with Nvidia.

• ▲ LOW DISSEMINATION — Eutelsat prepares 32 OneWeb renewal satellites for Nov. 30 SpaceX launch: Eutelsat’s OneWeb unit is set to launch 32 renewal satellites on a SpaceX vehicle slated for Nov 30【https://satnews.com/2026/11/12/eutelsat-prepares-32-oneweb-renewal-satellites-for-nov-30-spacex-launch/】. Desk take: Expands OneWeb’s LEO constellation, impacting broadband market share.

• ▲ LOW DISSEMINATION — SpaceX’s bid for Grain’s low‑band spectrum takes a step forward: The FCC cleared SpaceX’s application for Grain Management’s 800 MHz low‑band spectrum, a key asset for Starlink Mobile【https://www.lightreading.com/5g/spacex-s-bid-for-grain-s-lowband-spectrum-takes-a-step-forward】. Desk take: Securing low‑band spectrum could boost Starlink’s mobile connectivity offering.

• ▲ LOW DISSEMINATION — Salesforce releases SOV‑Transcribe‑1, now the top open‑weights speech model: Salesforce’s new SOV‑Transcribe‑1 model has become the leading open‑weights speech model according to Crypto Briefing【https://news.google.com/rss/articles/CBMiiAFBVV95cUxNTzQ4dmRNNG5hazV6bjRrU0Y4eWJrcFFkX2V5RnpGcTM2b1h0NDRfbDJOMkV0Q1gzRjRCV1VKTG1KNjRzc2R4N0JucDFmdF9iU1gta2xBZk5wdG01QnpjRXlteVJ4NHM3cmx4TEw0aUZYUWFWdk1KNjV0UW9tazcxbXFhOE1lMHYt?oc=5】. Desk take: Highlights Salesforce’s push into generative AI speech, a niche with monetization upside.

• ▲ LOW DISSEMINATION — Concentrate AI launches HIPAA‑compliant AI gateway with single BAA to simplify model access and cut inference costs: The new gateway offers a unified Business Associate Agreement for HIPAA compliance, easing healthcare AI deployment【https://news.google.com/rss/articles/CBMi-wFBVV95cUxQYTF3Z21CLWpjZUxQWkFjZmxpc2lzc1NrSjg5LTBuQk1nQ2t1bmlwXzJMQ2hHQXZGTDBsclFIVW9KTFpMRjgtaFd1amcxblA3MUtMeGM0TTNJNHU5ejF5a1BWQ3l1NVFZMlUzdmhlUHZOaHRLY1licjBDbjlrSHA0d2RqRnEyREtIcmQtOTA4N1hOZjBVRDVFVUF1WXlOWUZHNnVoMEgzbnZFUWxXTi1mcHQ1LTBLS1BXcUdUVWV0Ylg1S1BBakF4eFNSOWdpOTg0b1F4R2pxQjN5RE5oVUpWLURLQUhDMDFsVktMU1ljOUhYM3dGTkVzVGtWdw?oc=5】. Desk take: Could accelerate AI adoption in regulated health sectors.

• ▲ LOW DISSEMINATION — Tether launches on‑device AI agent BotPsy, claims win over larger Qwen as Qwen’s own score narrows: Tether’s BotPsy AI agent is positioned as outperforming larger Qwen models despite recent score improvements by Qwen【https://news.google.com/rss/articles/CBMi5AFBVV95cUxNNE5ZNGlwajhVcF9zemlhR21NNXEyZy1PeVRIcU9nbGloTGlhTmkxWEZCS2lvRmwzSjNVc3RpZzYxdk1WQVpaZkNpUXdwVld6RkRjcDVWU0R0b3A5R1BFeDduOEx3YkVTU29qWlVTdm43Z1JKbnlGU2haWXoxT0dCSnZuY2RBNmlGSnNmMTluMFhYc0d3UjdXSUdzbUJYQ2wzUEhrYk8xdkxGSENwYXdJM0lUOGRyNkFmb1lsQ19ja2dGMkNNYlc5dk0zY0VwU3BWZTdlaEV5NGd2eE5fZmExemZ6YmQ?oc=5】. Desk take: Signals growing competition in on‑device LLM space, relevant for edge AI strategies.

• ▲ LOW DISSEMINATION — Techie uses voltage glitching hack to turn $2,500 Nvidia 8GB cryptomining GPU into a full‑fat $20,000 A100: By blowing onboard fuses, a miner upgraded a CMP‑170HX 8 GB card to match an A100 80 GB’s performance【https://www.tomshardware.com/pc-components/gpus/techie-turns-usd2-500-cmp-170hx-8gb-mining-card-into-a-full-fat-a100-80gb-with-voltage-glitching-carefully-blowing-onboard-fuses-to-unlock-entire-vram-and-shader-modules】. Desk take: Highlights hardware vulnerability that could affect Nvidia’s product segmentation.

• ▲ LOW DISSEMINATION — Society’s four fiber transitions: Lightwave Online outlines four key shifts in fiber infrastructure, detailing emerging trends and deployment strategies【https://news.google.com/rss/articles/CBMijgFBVV95cUxPa3RIdUtYRzlTTnJlb0RTZklJc1hXS0Q1aEV0T2NVd1BrekdGeXg1dXVLT2ZNUUg2aWRjczEwTGl4dlBpM1loX1Q2bXNEVGszdXhIZWNHVTZhZmVQVW15Tkl3UEhJYVM4Zm1sallUejE3amFDN1hfMlhGQmVBNHE5MmVYbVpMN1RDc3dscnd3?oc=5】. Desk take: Provides insight into fiber market dynamics affecting telecom and data‑center connectivity.

• ▲ LOW DISSEMINATION — Kratos completes SRR for US Space Force missile‑warning programme: Kratos finished a System Requirements Review for a missile‑warning system under the US Space Force contract【https://news.google.com/rss/articles/CBMie0FVX3lxTFBITTUtRTFZNW8zSFhiN3daeGV3VWtTLUhaTFBjM0VyLTVMQnNzZlA1cG53LW9TaDdPRE1Gc0wwS2hkWkM4bVQ5ZDdhTHpQalNsNzR4Q05YRU03RldQM0RCUEY2bHN3Q0pjMkczN1lhQm5BQ2JrdURJLWFfcw?oc=5】. Desk take: Reinforces Kratos’s role in national security space contracts.

• ▲ LOW DISSEMINATION — Nvidia: Ex‑prosecutor Joon Kim and Wachtell, Lipton bring big track record to investigation: Nvidia has retained former prosecutor Joon Kim and law firm Wachtell, Lipton to lead its internal probe amid regulatory scrutiny【https://www.cnbc.com/2026/11/12/nvidia-jensen-huang-joon-kim-wachtell-china-spy.html】. Desk take: Legal developments could impact Nvidia’s governance and market perception.

• ▲ LOW DISSEMINATION — Lex Fridman article on Nvidia’s Jensen Huang: A Taiwanese tech news piece discusses Jensen Huang’s recent statements and Nvidia’s AI industry positioning【https://finance.technews.tw/2026/11/13/lex-fridman-jensen-huang/】. Desk take: Reflects ongoing narrative around Nvidia’s AI leadership in Asian markets.

• ▲ LOW DISSEMINATION — RAVEL and Sandia to present Project Quicksilver at SC26: The project uses RAVEL Orchestrate AI to explore distributed workload management, energy efficiency, and infrastructure economics【https://www.hpcwire.com/off-the-wire/ravel-and-sandia-to-present-project-quicksilver-at-sc26/】. Desk take: Could influence future HPC and AI workload orchestration strategies.

• ▲ LOW DISSEMINATION — Korea Investment cuts Naver target; says rebound hinges on AI data centers: The fund lowered its valuation for Naver, emphasizing that recovery depends on AI‑focused data‑center investments【https://news.google.com/rss/articles/CBMigAFBVV95cUxQNzFJUVNmVUFhQ01US0hialNOeW1qLXROaTBzbTZvT0dmcjhOUkZyTDFRbGxkbTg2X3dDNndWVUI4SXdNSjk0TC14UEdEeEgzaFl5cU9GSHFPSjdUaWZETmQtaGstMjg5TlM3

Conversational Intelligence

• swyx, host (Latent Space, 2026‑11‑07): Anthropic's compute spend for serving models has increased 2.3× year‑over‑year while revenue has risen 14×, implying a 7× efficiency gain in inference serving. “their compute that went towards serving models has gone up by 2.3 X their revenue over the same period of time has gone up 14 X so embedded in that is basically a 7 X efficiency gain in serving inference” [min 0]. Tag: Data unavailable — dissemination tag. Desk take: Significant inference cost improvements boost Anthropic’s margin outlook.

• swyx, host (Latent Space, 2026‑11‑07): Anthropic has raised $141 billion of capital to date and holds a $134 billion cash balance. “my favorite stat of the whole S1 was $141 billion of capital raised to date. That's inclusive of equity, converts, and safes, and a $134 billion cash balance today” [min 0]. Tag: Data unavailable — dissemination tag. Desk take: Massive cash cushion underpins Anthropic’s strategic flexibility.

• Jamin Ball, guest (Latent Space, 2026‑11‑07): Anthropic has raised $141 billion of capital to date, inclusive of equity, converts, and SAFEs. “$141 billion of capital raised to date. That's inclusive of equity, converts, and safes.” [min 15]. Tag: Data unavailable — dissemination tag. Desk take: Reinforces the capital strength highlighted by the host.

• Zeno Rocha, founder, Resend (Latent Space, 2026‑11‑10): Resend is powering 4.3 million users as of this morning. “Resend now is powering 4.3 million users as of this morning.” [min 2]. Tag: Data unavailable — dissemination tag. Desk take: Strong user traction signals growth potential for the email infrastructure play.

• unidentified (All‑In, 2026‑11‑07): Data center developer Shane said there is a 71% chance of going public this year, down from 95% before the midterms. “they're going to have a 71% chance of going public this year. It was 95% before the midterms.” [min 54]. Tag: Data unavailable — dissemination tag. Desk take: Diminished IPO odds may delay capital inflows for the developer.

• unidentified (All‑In, 2026‑11‑07): Approximately $4 trillion in market capitalization was wiped out in a single day across major AI and semiconductor stocks. “Close to $4 trillion in market cap wiped out.” [min 52]. Tag: Data unavailable — dissemination tag. Desk take: Broad equity correction heightens valuation risk in the AI sector.

• David Friedberg, host (All‑In, 2026‑11‑07): The venture firm invests in 100 companies a year, spends about $40‑$50k on AI subscriptions, estimates $2 million of value, $1 million incremental revenue and $15‑$21k incremental revenue per dollar spent. “I invest in 100 companies a year... we're spending like $50,000, I think, $40,000, $50,000 on these subscriptions... I estimate we're getting $2 million in value... about $1 million in incremental revenue... maybe $15,000 to $21,000 in incremental revenue for the dollars we spend.” [min 72]. Tag: Data unavailable — dissemination tag. Desk take: High ROI on AI tools supports continued vendor spend.

• Rick Smith, CEO, Axon (All‑In, 2026‑11‑10): Axon owed Silicon Valley Bank about $2 million. “we owe Silicon Valley Bank about $2 million” [min 8]. Tag: Data unavailable — dissemination tag. Desk take: Outstanding SVB debt adds a modest liability to Axon’s balance sheet.

• unidentified (Moonshots, 2026‑11‑07): Elon Musk has said there is roughly 15 gigawatts of idle compute capacity that cannot find a data center with liquid cooling to use. “Elon has been saying for a while that there's going to be about 15 gigawatts of idle compute capability that can't find a data center with liquid cooling to plug into.” [min 28]. Tag: Data unavailable — dissemination tag. Desk take: Untapped compute could become a supply source if cooling constraints are solved.

• unidentified (Moonshots, 2026‑11‑07): In a best‑case scenario, a Dyson Swarm could capture a couple percent of all global compute within two years. “best case that's a couple percent of all compute in two years and then it's on an exponential so it could really take off after that” [min 28]. Tag: Data unavailable — dissemination tag. Desk take: Space‑based compute concepts may soon impact total compute capacity.

• unidentified (Moonshots, 2026‑11‑07): Artificial Superintelligence (ASI) is expected to emerge within a two‑year window. “we all know that ASI is within that two-year timeline.” [min 28]. Tag: Data unavailable — dissemination tag. Desk take: Near‑term ASI expectations could drive speculative investment flows.

• Alex (Moonshots, 2026‑11‑07): Researchers estimate that roughly 80% of the acceleration is due to AI‑assisting AI research. “So researchers we're estimating that this is roughly 80% of this acceleration is AI‑assisting AI research.” [min 38]. Tag: Data unavailable — dissemination tag. Desk take: AI‑in‑AI feedback loops are a key driver of rapid progress.

Tape Check

Snapshot as of 2026-11-13 08:40 UTC.

Today / Watch

AI‑centric hardware and memory breakthroughs are reshaping space infrastructure and data‑center capacity amid rising geopolitical constraints

115 stories were gathered and ranked this morning; the full ranked listing appears in The Full Tape below.