Thursday, October 8, 2026

Back to today

Boyz in the Hood

Robinhood’s AI agents, 24/7 equity trading and 10:1 crypto perpetual futures encourage higher retail trading velocity, leverage and momentum.

From X

@StableBread 4 clicks

Fajasy on X

One of the world’s 10 biggest financial institutions signed a deal in 2024 to deploy software from Nubeva Technologies $NBVAF (TSXV: $NBVA). Nubeva won’t name the bank, but its filings leave a trail. The April 2024 release on the deal said revenue from the first 12 months “is not expected to be material,” and no press release or filing has updated the deal since. Nubeva reports revenue by country, and in FY2026, U.K. revenue was $204,055, the same figure as the license line. Nubeva also discloses what share of revenue its largest customers make up. In both of the last two years, U.K. revenue matched one customer’s share almost exactly. In FY2026, U.K. revenue was 40.8% of the total ($204,055 / $500,586), and the two largest customers were 41% each. In FY2025, U.K. revenue was 36.6% ($193,986 / $530,616), and the second-largest customer was 37%. One of the two 41% customers lines up with the support and maintenance line, 41.1% of FY2026 revenue ($205,712 / $500,586), which the last customer of Nubeva’s old TLS (web-traffic decryption) product pays. That means the license revenue, the U.K. revenue, and the remaining 41% customer all appear to be the same account. The FY2025 statements show the 37% customer at 0% of FY2024 revenue, so it started paying in FY2025, the first full fiscal year after the bank deal. In a June 2024 video, CEO Randy Chou said the bank doesn’t want anyone to know it uses Nubeva. If the U.K. customer is the bank, HSBC $HSBC is the likeliest candidate. It’s based in London, and at eighth by assets, it’s the only U.K.-headquartered bank in S&P Global $SPGI’s top 10. Then there’s Netskope $NTSK. The company that bought Nubeva’s key-capture patents runs a public page titled “Netskope + HSBC” that pitches its platform to the bank. It’s a sales page, with a “Meet with your Netskope team” button and a named Netskope sales rep, and it pitches Netskope against a “Bluecoat / Zscaler hybrid approach,” which reads like HSBC’s current setup. It also shows Netskope courting HSBC, not HSBC buying from Netskope, and nothing on the page mentions Nubeva or ransomware. If the bank is HSBC, Netskope, which owns the key-capture patents Nubeva sold, is already pitching Nubeva’s biggest license customer. None of this proves the bank is HSBC. But in the latest quarter (February-April 2026), instead of growing, revenue from the U.K. customer fell ~70% (vs. a 19% rise in the same quarter a year earlier). License revenue was $18,796 ($204,055 for the year minus $185,259 through January), vs. the ~$62K/quarter it averaged in the first three quarters. The FY2026 MD&A doesn’t say whether the license customer renewed. If it didn’t, Nubeva likely lost the bank (possibly HSBC) behind 41% of its FY2026 revenue. The six-month report due by December 30, 2026 will show whether license revenue came back.
@convequity

Convequity on X

$RBRK's backup engine breaks data into fixed-size blocks, for example 4 MB, and stores only the unique blocks across backups. Commvault and Cohesity instead divide data into variable-sized chunks, with the boundaries set by the content itself. Each design has a trade-off. With fixed blocks, a small change inside a block means the whole block gets written again, so Rubrik uses more storage. In return, hashing and comparing blocks is simple, which means faster backups and lower compute overhead. With variable chunks, only the changed portions are written, so storage use is much lower. Working out chunk boundaries and hashes as the data comes in takes more CPU and memory, which can slowdown backups. Cohesity offsets part of this with a scale-out architecture that processes in parallel across nodes. The two designs fit two different jobs. Variable-length deduplication suits long-term archives, where storage cost matters more than speed. Fixed-block deduplication suits recovery after a hack or ransomware attack, where fast restores and predictable performance matter more than saving storage. That matches how each company started. Rubrik came into the market in 2014 focused on security and on getting critical data back quickly after a breach. Cohesity started from a different idea. It built its own file system that spreads data across many servers, each with its own computing power. Backup copies stay readable like normal files, so companies could search and analyze them where they sit, without restoring them first. Security came second. The two are now moving closer together. Cohesity has leaned much more into security and recovery, especially after buying Veritas's backup business in 2024.
@convequity

Convequity on X

Huawei's finer bonding deserves more weight in the AI-chip race. $AVGO's 3.5D approach combines vertically stacked dies with the side-by-side packaging used to connect processors and HBM. It adds another direction in which designers can shorten the wires moving data around a chip. Besi describes Broadcom's initial face-to-face hybrid bonding at 6µm pitch. Huawei reports 1.5µm for its 2026 Kirin implementation. Pitch is the spacing between connections; face-to-face brings the dies' wiring surfaces together. On an identical square grid, that spacing difference allows 16x as many potential bond sites per unit area. Usable connections still depend on routing and the space reserved for power and repair. It is an architectural opportunity, not a bandwidth multiple. The size qualification matters. Kurnal's Kirin 9050 die shots measure roughly 121mm² per die. Extending fine-pitch bonding to much larger AI dies changes the economics: more expensive silicon is at risk when a stack fails, and heat and power delivery have to work across the whole assembly. Broadcom also said face-to-face compute SoC shipments began in February 2026. A later roadmap cannot be read as its first arrival at face-to-face bonding. Huawei looks ahead on this published pitch comparison. Our view is that this gives it a credible route to narrowing parts of the AI-chip gap through packaging. The stronger claim needs production evidence at large accelerator scale. The phone result alone does not establish when that arrives. Builds on Convequity's Huawei After Sanctions, Parts 1–2, 10 & 18 Sep 2026, with updated source checks.

Golar LNG q2’26 Operational Update

Golar LNG ordered a $2.45 billion fourth FLNG vessel, targeting 3.5MTPA of capacity and $400 million in annual EBITDA under a 20-year…

Top Rebound Stocks for Oct’26

Intuitive Surgical, down 32% from its January 2025 peak, faces slower US procedure growth and new robotic-surgery rivals despite $10.1…

Vertiv Holdings Co. (VRT)

Vertiv's 33% stock decline followed a 2026 forecast increase, despite a $15 billion backlog and rising AI data-center demand for power and…

POST · long

Post Holdings INC

Post Holdings remains a large position after the stock's recent decline.

What I’ve Been Up To

Cogent Communications and AI data-center bottlenecks feature in a newsletter update on recruiting, financial media and conference takeaways.

Publications we follow

Every publication on this site, hand-picked.

109 publications

Newsletters 92

X accounts we follow

Every account on this site, hand-picked.

74 accounts

Investor letters we follow

189 firms

Sectors

Ordered by how often our editor reaches for them.