Dmytro Lebid

I didn’t like $SIVE Q2 2026 report, which reveals dangerous cracks in the fundamentals. The harsh reality of the numbers turned out to be more important than the hype surrounding their AI photonics. Certainly, we received positive news about the growth of the pipeline portfolio to $1.2B. However, the operating picture doesn’t come close to reflecting that. Revenue fell by 12% YoY, while the net loss increased 2.3-fold. At the same time, the negative operating cash flow balance is growing. On top of that, the factors of additional share issuances and debt conversion—which dilute shareholders’ equity—are not going away. As a result, the break-even point in 2027 still carries the risk of a liquidity trap. I want confirmation in the form of recurring contracts that will generate actual revenue. As I understand it, the market shares my negative view, which is why Sivers’ stock continues to fall.

Goldman Sachs forecasts that AI CapEx will double by 2031, rising from $765B to $1.64T. And this is despite the fact that, based on their own projections, FCF for hyperscalers will double as early as 2029. Global spending on AI infrastructure will surpass the $1T mark as early as next year. Spending on silicon and computing clusters will play a dominant role (69% of all budgets). Investments in data centers will increase to $436B. Power capacity, however, is modestly estimated at $73B in CapEx. I believe we shouldn’t be swayed by skeptical claims about an imminent contraction in capital expenditures. Momentum has built up significantly, and chips ( $NVDA, $INTC, $AMD ) will continue to account for the lion’s share of AI CapEx revenue.

$NBIS announced its intention to issue $4.5B in convertible bonds. Against this backdrop, its stock price is falling. But let’s consider how much this money is actually needed given the current circumstances. Currently, NBIS has $8B in liquidity reserves. However, projected CapEx for the next quarter is $8.7B. NBIS reported that 50–60% of CapEx is financed by prepayments. However, this will still not be enough to last more than two quarters. After all, CapEx totaled $5.7B in the reporting quarter. And in 2027, it will most likely continue to increase. I believe that the issuance of convertible bonds should not be viewed negatively. The company will use the proceeds to further scale its AI cloud platform, expanding its contracts to achieve its corporate goals. This is a necessity. By doing so, the company is paying with its market capitalization. But this does not change its global mission or negate its long-term goals of becoming the best cloud platform.

Wall Street analysts’ median price target for $NBIS is $272.56, and I believe this is an underestimate given that Nebius still has 3 GW of uncommitted capacity. We have learned that short-term contracts of up to 3 months are being signed at a rate of $40–50M per 1 MW. This is a trigger for parabolic growth in profits and business margins. However, there are institutional investors who share my view on NBIS’s enormous potential. Analysts at Northland Securities have raised their price target to $410, while analysts at Citi have raised theirs to $324. Analysts at Bank of America Securities have set a price target of $310 with a “Buy” rating. Even analysts at D.A. Davidson, with their “Hold” rating, raised their price target from $175 to $250. But they may be wrong, as NBIS’s potential has not yet been fully realized.

$POET quarterly report confirms the ongoing transition from an R&D project to the commercialization phase of its AI-photonics technologies. We’ve seen several positive signs: Revenue grew by 112% YoY (up 13% QoQ)—marking 6 consecutive quarters of sustained growth. The net loss narrowed to -$0.07 per share. Cash reserves reached an impressive $796.3 million. The commercial pipeline is taking off: an agreement with Lumilens with a potential of $500+ million over 5 years and the start of serial shipments for qualification in 2H 2026. My conclusion: POET’s valuation reflects its current modest revenue. But once the $800 million balance sheet is prudently allocated and the company’s readiness to scale up as it commercializes products under the Lumilens agreement is confirmed, we can expect a significant revaluation. But I’m not sure this will happen before 2027. What do you think—will POET’s hybrid integration allow it to capture its share of the market? Share your thoughts in the comments.👇

$SIVE has announced a new strategic move, revealing a $3.4M program with SemiNex to develop InP light sources. The main “bottleneck” in scaling AI clusters is not GPUs, but optical interconnects. This deal represents yet another attempt to bring a viable product to market as a CPO for data centers. Energy efficiency and thermal stability of lasers are key factors in bringing optics onto the chip. SIVE plans to launch pilot production of the product in 2H 2027, setting the stage for revenue growth starting in 2028. I believe that the market still views SIVE as a niche supplier but is overlooking its transformation into an independent architect of the optical stack. We are seeing yet another confirmation that SIVE’s long-term upside is enormous.

$NBIS and $IREN are on the verge of a potential short squeeze, and there is direct evidence to support this. The potential for a short squeeze across the sector as a whole continues to fuel a potential surge in stock prices. Take a look at the chart below, which isn’t just a list of companies, but a map of the market’s potential fuel tank. My thesis is that the higher the short interest in the cloud AI infrastructure sector, the greater the likelihood of a built-up spring effect during a short squeeze. The fact that NBIS has a short interest of 27.71% is not a sell signal. Despite enormous pressure from bears who believe that AI CapEx growth won’t pay off, even the slightest positive shifts in revenue will trigger a chain reaction. When more than 20–25% of sales are forced to exit through a single narrow exit, the price skyrockets parabolically. I also include IREN shares in this category, which have a short interest of 23.97%. NBIS and IREN are holdings in my portfolio, and I disagree with the market’s extreme pessimism toward these companies. The fundamental race for capacity starting next year will begin to drive explosive revenue growth, setting the stage for a future short squeeze.

Great news for $USAR investors. The U.S. has announced plans to develop critical mineral resources with an additional $2.18B in funding. The White House announcement confirms a fundamental turning point for the rare earth metals sector. The planned government investments reduce systemic risks for USA Rare Earth, which receives not just capital, but a guarantee of demand in the domestic market. Given the ongoing strict protectionist measures, the establishment of a closed-loop rare earth metals cycle within the U.S. is accelerating USAR’s market capitalization. I believe that USAR is transforming from a cyclical mining company into an indispensable infrastructure asset. And due to the shift in focus toward national security, we will see a multiplier effect in its valuation.

Great news for $USAR investors. The U.S. has announced plans to develop critical mineral resources with an additional $2.18B in funding. The White House announcement confirms a fundamental turning point for the rare earth metals sector. The planned government investments reduce systemic risks for USA Rare Earth, which receives not just capital, but a guarantee of demand in the domestic market. Given the ongoing strict protectionist measures, the establishment of a closed-loop rare earth metals cycle within the U.S. is accelerating USAR’s market capitalization. I believe that USAR is transforming from a cyclical mining company into an indispensable infrastructure asset. And due to the shift in focus toward national security, we will see a multiplier effect in its valuation.

Lumilens’ emergence from the shadows with up to $900 million in funding represents the primary institutional validation of $POET AI-photonics. The $5.51B valuation of POET’s strategic partner signals that investors are redirecting their billions in capital from pure compute toward overcoming the infrastructure bottleneck of inter-chip connections. Lumilens’ debut definitively confirms that architectures utilizing optical interposers are becoming the standard for CPO and NPO in AI data centers. But this also provides additional assurance for POET, with funding expected in the coming quarters for the fulfillment of a previously signed contract. Let’s not forget that POET possesses a strong patent portfolio that enables it to scale the growth of TAM. However, Lumilens’ valuation of $5.51B, in my opinion, reflects a deep discount relative to POET’s public-peer valuations. This is a clear bullish signal not only for POET but also for other players, including $SIVE. Moreover, Sivers supplies this market with key DFB lasers, creating a unified technological ecosystem with POET.