BREAKING: hundreds of NVIDIA employees have signed an internal letter demanding the reinstatement of Jensen Huang as CEO and the resignation of the board
@Lars_Hansen_86@MBrgger No he has not. It's just you that can't or comprehend. Russia has more than 1.5 million losses in Ukraine, that means killed AND wounded. A soldier with legs blown off can't fight anymore.
The 450.000 killed is quite accurate.
$SIVE has specific technical reasons to win new AI optics business after the $MRVL + $LITE deal.
The part I think deserves more attention is how @SiversSemicond designs, manufactures and integrates its lasers.
Start with the laser arrays.
$SIVE can combine multiple precisely controlled wavelengths in an array designed for silicon photonics.
For a customer building an optical engine, that can simplify assembly by reducing the number of individual components that need to be positioned and aligned.
Packaging is expensive.
A laser that performs well and makes the complete module easier to manufacture can be commercially attractive even before you discuss supply diversification.
Then there is the manufacturing process.
$SIVE uses etched facets, wafer-level coating and optical testing before the wafer is cut into individual chips.
According to the company, this eliminates bar processing and supports better yield, scalability and cost efficiency.
Those details matter when a customer needs millions of consistently performing devices.
Engineering a working laser and manufacturing it economically at volume require different capabilities.
$SIVE also offers flexibility across power, wavelength and form factor.
Its 70mW and 100mW lasers target applications including 800G and 1.6T pluggable transceivers, while its arrays address the integration requirements of optical I/O and co-packaged optics.
There is already concrete integration work behind the technology.
@AyarLabs and $SIVE announced a 16-wavelength laser-array demonstration within the SuperNova light source.
$GFS is incorporating $SIVE arrays into silicon photonics reference designs and its SCALE platform.
That gives prospective customers existing development work to evaluate as they select a light source for their own architecture.
$SIVE as also a founding member of the CW-WDM MSA alongside companies including $INTC, $LITE, $MTSI and @AyarLabs.
It has been involved in defining the specifications for these multi-wavelength light sources.
This is why I think $SIVE could attract more serious consideration now: a customer seeking independent supply can evaluate a company with relevant laser designs, manufacturing processes and established integration relationships.
The choice will still depend on performance, qualification, price and delivery.
But $SIVE has a credible technical case on each customer’s shortlist.
If that translates into production wins, I believe the commercial upside could be substantial.
For companies like $SIVE / $AAOI, the demand imbalance for lasers are so severe
That even a 20-30% overall demand reduction from a hypothetical DC slowdown would likely not eliminate the shortage.
We're seeing this before the next architectural shifts, where $NVDA Jensen Huang expects 15x optical volume for CPO scale-up.
And as more hyperscalers commit to NPO as seen with $LITE new $5B UHP laser contracts.
I'd argue this is more nuanced since I can't say the same for finished 800G/1.6T pluggables or lower end 70mW-100mW laser ranges that are currently bottlenecks.
But this argument is more toward higher spec lasers.
$AAOI is vertically integrated and has 300-400 mW capabilities, so they have the capability to shift attention to merchant lasers over supplying end-to-end optical transceivers.
$SIVE is more pure play on this laser bottleneck, but they have the capabilities for 200mW + higher margin multiwavelength NPO/CPO lasers.
I'd argue this is specifically the case for bottlenecks where demand imbalances are so deep (creating higher long term visibility, such as HBM).
Whether it's large body substrates from $ATS that's bottlenecking Marvell + $AMD or lasers from $LITE that are sold out into 2029.
or liquid transformers that have 4-5 year lead times.
They they would still sell out of capacity even if overall demand decreased.
If markets de-risked, it would make sense to hit optical module manufacturing players, maybe server rack assembly, to companies that still require heavy financing like Neoclouds.
And obviously haircut growth valuations (but not to the point where demand collapses to 0).
Yet we're seeing everything in AI trade is getting lumped up together as something that requires heavy financing + not profitable.
Russians have released drone footage of them dropping a multi-ton guided bomb on a school in Zaporizhzhia, setting it to pop music.
This strike killed over 30 people.
This is Russian culture.
Following the Lumentum / Marvell deal:
My opinion is that independent Western laser players, like $AAOI, $COHR, and $SIVE became a lot more strategically valuable.
$MTSI and $SMTC should get some attention too as emerging laser players in 2027.
If I had to guess:
1. Companies competing against Marvell, or not in the $NVDA ecosystem bandwagon, such as $AMD, are likely going to rethink their relationship with $LITE.
If getting bottlenecked by Nvidia capacity agreements wasn't enough... Incentives around qualifying + bringing up other laser players, seems very necessary now.
Since you generally don't want competitors like Marvell to have influence or visibility over your roadmaps, IP, capacity, pricing, etc. even if there is a "firewall".
2. $MRVL acquiring Lumentum opened pandoras box in a way of acquiring a critical chokepoint all your competitors + buyers use.
$LITE was used by your major hyperscalers and across AI supply chains.
Now, your leaders might wonder if weaponizing capacity agreements might not be as effective.... If someone else can just buy the entire company.
There's only a few major public merchant Western players out there, and I'm also sure many others will want to follow the Marvell / Broadcom route via vertical integration, given I think $MRVL saves around $1B+ in costs/year.
So M&A premiums likely went up.
$SMTC was actively looking to buy laser capacity via M&A so I'm sure they're looking at $SIVE given their 100M CW DFB laser expansion + tiny MC.
$JBL also probably wants to vertically integrate down the road too so $SIVE is also an acquisition target for their 1.6T program.
$AMD was looking for ELSFP + CW capacity given Nvidia bottlenecks, so I'm sure they've looked at $AAOI.
can go on and on about other ASIC players (like in Taiwan) that likely are looking at vertical integration too.
Since multiple players use a small subset of laser suppliers, if competitors just go ahead and buys them...
There's a lot pressure to M&A for their own supply chain defensibility.
TLDR: Independent laser suppliers should capture a material strategic revaluation since $MRVL + $LITE, sets a new precedent + implications.
We'll see what happens.
"They killed us." With these words Novaya Gazeta, Russia's most famous independent newspaper, announced it is closing. Six of its people were murdered under Putin.
🧵Here is what Russia is losing. [1/12]
@BTFDGreg@CoinGatecom I know. Was just referring to exchange volume regarding that.
As for LTC having more txs, yes, that's correct. However, it can't scale, so if real crypto adoption picks up it will start to fail.
@BTFDGreg@CoinGatecom There are people using it. But no fake volume & very little wash trading.
BCH has ~15k - 20k tx pr day. But you're right, the upside is huge if it takes just 10% of the overall tx volume in crypto. I don't think many people realize how many top 20 coins fail at quantum & scale.
@BTFDGreg@CoinGatecom On the contrary. The need for a censorship-free p2p cash, DeFi capable and (real) scaling and quantum-safe crypto is still real and BCH has exactly that.
MORE RUSSIAN TERRORISM: A Russian jet-powered drone struck a crowded railway platform in Dnipro, Ukraine, killing at least 14 people and injuring 14 others.
Two passenger trains were at the station, including one preparing to depart for Zaporizhzhia. Ukraine’s Air Force said it intercepted three drones, but a fourth got through and hit the station.
The attack came just hours after Trump claimed there was “a lot of progress” in talks to end the war, adding that a meeting between Putin and Zelensky would be set up “very soon.”
@BTFDGreg@CoinGatecom Sure. I think you underestimate BCH. After all it's now part of most institutional instruments and categorized as a commodoty like BTC - as the only clean fork mind. It's technically superior to BTC and EVMs, so the upside is much larger than the downside.