Technology Newsround – Adobe & Cambricon

Adobe Q1 26 – Unnecessary uncertainty

• A good set of results was spoiled by an injection of uncertainty, as Adobe CEO Shantanu Narayen has “resigned” as CEO, with no real explanation, giving the impression that something is wrong when the numbers point to precisely the opposite.
• FQ1 26 revenues / Adj-EPS were $6.40bn (+12% YoY) / $6.06, nicely ahead of consensus of $6.28bn / $5.87, and guidance was raised slightly for the coming quarter.
• FQ2 26 revenues / Adj-EPS are expected to be $6.46bn ($6.43bn – $6.48bn) / $5.83 ($5.80 – $5.85) again just ahead of consensus of $6.43bn / $5.70.
• Everything is moving in the right direction with AI-first annualised revenue tripling (albeit from a small base), the highest level of cash flow from operations ever recorded, and a diluted share count that fell to 411m from 417m at the end of Q4 25 and 438m at the end of Q1 25.
• Into this peaceful progress has been injected an agent of uncertainty as the CEO is moving on, but will remain in place until a successor is found.
• Mr Narayen will remain chairman of the board to support the new CEO.
• I am far from convinced that Adobe needs the change of direction that a new CEO is likely to bring, as it has already done what it needs to do to address the advent of AI, and simply has to keep executing.
• I fear that this is a knee-jerk reaction to the collapse in the share price, which I continue to think has very little to do with Adobe and much more to do with a sector-wide derating of earnings due to AI fears.
• The other problem is that his departure is being badly handled, and investors are none the wiser as to the real reason for his “resignation”.
• This creates uncertainty, which the market hates, explaining the decline in the share price when all the company really needs to do is more of the same.
• Hence, I hope that this is what the new CEO decides is the best course of action, and so I am looking for a very boring appointment.  
• With the shares down 7% after-hours and 2% below where I bought them, I remain happy with my position but would look to top up if there is further sustained weakness.

Cambricon – A rounding error.

• Nvidia’s No. 2 challenger in China reported a profit for 2025, but even with its intention to triple its output, it will barely make a dent, underlying that China has a big problem with compute production as well as performance.
• In 2025, Cambricon recorded revenues of RMB6.5bn ($944m) and net income of RMB2.1bn ($306m), representing a good improvement over 2024.
• Cambricon has also said that it intends to triple its output in 2026, which would bring chip output to 300,000 AI accelerators costing around $10,000 each.
• Like Huawei, Cambricon is using the 7-year-old 7nm multipattering process that is falling further and further behind the leading edge with every passing generation.
• If I assume that Cambricon’s accelerators are roughly the same as the Ascend 950 (Huawei’s 2026 chip), then its entire output for 2026 will offer around 600 ExaFLOPS of performance.
• This is compared to Huawei, which RFM estimates will produce around 8,000 ExaFLOPS in total in 2026, and Nvidia, which RFM estimates will produce 229,500 ExaFLOPS in 2026.
• This means that Nvidia will produce 26.7x more compute in 2026 than both of these two companies combined, highlighting that China has a problem not just with the performance of its chips but also in the total volume of compute that it can produce.
• This is why I have serious doubts with regard to China’s ability to roll out AI services at scale, meaning that it will remain dependent on data centres in countries like Malaysia, Thailand and so on.
• These countries do not have the same restrictions and are able to build data centres using Western chips just like anyone else.
• This represents a significant workaround to the chip restrictions and one that I expect to be addressed when the US Department of Commerce next updates its rules in Q3 or Q4 of this year.
• Hence, I see no real sign that China is in a position to produce anything like enough compute to satisfy its demands, and with no foreign imports to rely on, it is likely to fall further and further behind.
• In the great struggle for supremacy in AI, I continue to think that the West and the USA in particular still have the advantage.